501 U.S. 452 (1991)

CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE EIGHTH CIRCUIT

[454] Jim J. Shoemake argued the cause for petitioners. With him on the briefs were Thomas J. Guilfoil and Bruce Dayton Livingston.

James B. Deutsch, Deputy Attorney General of Missouri, argued the cause for respondent. With him on the brief were William L. Webster, Attorney General, and Michael L. Boicourt, Assistant Attorney General.^[1]

Article V, § 26, of the Missouri Constitution provides that "[a]ll judges other than municipal judges shall retire at the age of seventy years." We consider whether this mandatory retirement provision violates the federal Age Discrimination in Employment Act of 1967 (ADEA or Act), 81 Stat. 602, as amended, 29 U. S. C. §§ 621-634, and whether it comports with the federal constitutional prescription of equal protection of the laws.

Petitioners are Missouri state judges. Judge Ellis Gregory, Jr., is an associate circuit judge for the Twenty-first Judicial Circuit. Judge Anthony P. Nugent, Jr., is a judge of the Missouri Court of Appeals, Western District. Both are subject to the § 26 mandatory retirement provision. Petitioners were appointed to office by the Governor of Missouri, pursuant to the Missouri Non-Partisan Court Plan, Mo. Const., Art. V, §§25(a)-25(g). Each has, since his appointment, been retained in office by means of a retention election in which the judge ran unopposed, subject only to a "yes or no" vote. See Mo. Const., Art. V, §25(c)(1).

[456] Petitioners and two other state judges filed suit against John D. Ashcroft, the Governor of Missouri, in the United States District Court for the Eastern District of Missouri, challenging the validity of the mandatory retirement provision. The judges alleged that the provision violated both the ADEA and the Equal Protection Clause of the Fourteenth Amendment to the United States Constitution. The Governor filed a motion to dismiss.

The District Court granted the motion, holding that Missouri's appointed judges are not protected by the ADEA because they are "appointees. . . `on a policymaking level'" and therefore are excluded from the Act's definition of "employee." App. to Pet. for Cert. 22. The court held also that the mandatory retirement provision does not violate the Equal Protection Clause because there is a rational basis for the distinction between judges and other state officials to whom no mandatory retirement age applies. Id., at 23.

The United States Court of Appeals for the Eighth Circuit affirmed the dismissal. 898 F. 2d 598 (1990). That court also held that appointed judges are "`appointee[s] on the policymaking level,'" and are therefore not covered under the ADEA. Id., at 604. The Court of Appeals held as well that Missouri had a rational basis for distinguishing judges who had reached the age of 70 from those who had not. Id., at 606.

We granted certiorari on both the ADEA and equal protection questions, 498 U. S. 979 (1990), and now affirm.

The ADEA makes it unlawful for an "employer" "to discharge any individual" who is at least 40 years old "because of such individual's age." 29 U. S. C. §§ 623(a), 631(a). The term "employer" is defined to include "a State or political subdivision of a State." § 630(b)(2). Petitioners work for the State of Missouri. They contend that the Missouri [457] mandatory retirement requirement for judges violates the ADEA.

As every schoolchild learns, our Constitution establishes a system of dual sovereignty between the States and the Federal Government. This Court also has recognized this fundamental principle. In Tafflin v. Levitt, 493 U. S. 455, 458 (1990), "[w]e beg[a]n with the axiom that, under our federal system, the States possess sovereignty concurrent with that of the Federal Government, subject only to limitations imposed by the Supremacy Clause." Over 120 years ago, the Court described the constitutional scheme of dual sovereigns:

"`[T]he people of each State compose a State, having its own government, and endowed with all the functions essential to separate and independent existence,' . . . `[W]ithout the States in union, there could be no such political body as the United States.' Not only, therefore, can there be no loss of separate and independent autonomy to the States, through their union under the Constitution, but it may be not unreasonably said that the preservation of the States, and the maintenance of their governments, are as much within the design and care of the Constitution as the preservation of the Union and the maintenance of the National government. The Constitution, in all its provisions, looks to an indestructible Union, composed of indestructible States." Texas v. White, 7 Wall. 700, 725 (1869), quoting Lane County v. Oregon, 7 Wall. 71, 76 (1869).

The Constitution created a Federal Government of limited powers. "The powers not delegated to the United States by the Constitution, nor prohibited by it to the States, are reserved to the States respectively, or to the people." U. S. Const., Amdt. 10. The States thus retain substantial sovereign authority under our constitutional system. As James Madison put it:

[458] "The powers delegated by the proposed Constitution to the federal government are few and defined. Those which are to remain in the State governments are numerous and indefinite. . . . The powers reserved to the several States will extend to all the objects which, in the ordinary course of affairs, concern the lives, liberties, and properties of the people, and the internal order, improvement, and prosperity of the State." The Federalist No. 45, pp. 292-293 (C. Rossiter ed. 1961).

This federalist structure of joint sovereigns preserves to the people numerous advantages. It assures a decentralized government that will be more sensitive to the diverse needs of a heterogenous society; it increases opportunity for citizen involvement in democratic processes; it allows for more innovation and experimentation in government; and it makes government more responsive by putting the States in competition for a mobile citizenry. See generally McConnell, Federalism: Evaluating the Founders' Design, 54 U. Chi. L. Rev. 1484, 1491-1511 (1987); Merritt, The Guarantee Clause and State Autonomy: Federalism for a Third Century, 88 Colum. L. Rev. 1, 3-10 (1988).

Perhaps the principal benefit of the federalist system is a check on abuses of government power. "The `constitutionally mandated balance of power' between the States and the Federal Government was adopted by the Framers to ensure the protection of `our fundamental liberties.'" Atascadero State Hospital v. Scanlon, 473 U. S. 234, 242 (1985), quoting Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S. 528, 572 (1985) (Powell, J., dissenting). Just as the separation and independence of the coordinate branches of the Federal Government serve to prevent the accumulation of excessive power in any one branch, a healthy balance of power between the States and the Federal Government will reduce the risk of tyranny and abuse from either front. Alexander Hamilton explained to the people of New York, perhaps optimistically, that the new federalist system would [459] suppress completely "the attempts of the government to establish a tyranny":

"[I]n a confederacy the people, without exaggeration, may be said to be entirely the masters of their own fate. Power being almost always the rival of power, the general government will at all times stand ready to check the usurpations of the state governments, and these will have the same disposition towards the general government. The people, by throwing themselves into either scale, will infallibly make it preponderate. If their rights are invaded by either, they can make use of the other as the instrument of redress." The Federalist No. 28, pp. 180-181 (C. Rossiter ed. 1961).

James Madison made much the same point:

"In a single republic, all the power surrendered by the people is submitted to the administration of a single government; and the usurpations are guarded against by a division of the government into distinct and separate departments. In the compound republic of America, the power surrendered by the people is first divided between two distinct governments, and then the portion allotted to each subdivided among distinct and separate departments. Hence a double security arises to the rights of the people. The different governments will control each other, at the same time that each will be controlled by itself." Id., No. 51, p. 323.

One fairly can dispute whether our federalist system has been quite as successful in checking government abuse as Hamilton promised, but there is no doubt about the design. If this "double security" is to be effective, there must be a proper balance between the States and the Federal Government. These twin powers will act as mutual restraints only if both are credible. In the tension between federal and state power lies the promise of liberty.

[460] The Federal Government holds a decided advantage in this delicate balance: the Supremacy Clause. U. S. Const., Art. VI, cl. 2. As long as it is acting within the powers granted it under the Constitution, Congress may impose its will on the States. Congress may legislate in areas traditionally regulated by the States. This is an extraordinary power in a federalist system. It is a power that we must assume Congress does not exercise lightly.

The present case concerns a state constitutional provision through which the people of Missouri establish a qualification for those who sit as their judges. This provision goes beyond an area traditionally regulated by the States; it is a decision of the most fundamental sort for a sovereign entity. Through the structure of its government, and the character of those who exercise government authority, a State defines itself as a sovereign. "It is obviously essential to the independence of the States, and to their peace and tranquility, that their power to prescribe the qualifications of their own officers . . . should be exclusive, and free from external interference, except so far as plainly provided by the Constitution of the United States." Taylor v. Beckham, 178 U. S. 548, 570-571 (1900). See also Boyd v. Nebraska ex rel. Thayer, 143 U. S. 135, 161 (1892) ("Each State has the power to prescribe the qualifications of its officers and the manner in which they shall be chosen").

Congressional interference with this decision of the people of Missouri, defining their constitutional officers, would upset the usual constitutional balance of federal and state powers. For this reason, "it is incumbent upon the federal courts to be certain of Congress' intent before finding that federal law overrides" this balance. Atascadero, supra, at 243. We explained recently:

"[I]f Congress intends to alter the `usual constitutional balance between the States and the Federal Government,' it must make its intention to do so `unmistakably clear in the language of the statute.' Atascadero [461] State Hospital v. Scanlon, 473 U. S. 234, 242 (1985); see also Pennhurst State School and Hospital v. Halderman, 465 U. S. 89, 99 (1984). Atascadero was an Eleventh Amendment case, but a similar approach is applied in other contexts. Congress should make its intention `clear and manifest' if it intends to pre-empt the historic powers of the States, Rice v. Santa Fe Elevator Corp., 331 U. S. 218, 230 (1947) . . . . `In traditionally sensitive areas, such as legislation affecting the federal balance, the requirement of clear statement assures that the legislature has in fact faced, and intended to bring into issue, the critical matters involved in the judicial decision.' United States v. Bass, 404 U. S. 336, 349 (1971)." Will v. Michigan Dept. of State Police, 491 U. S. 58, 65 (1989).

This plain statement rule is nothing more than an acknowledgment that the States retain substantial sovereign powers under our constitutional scheme, powers with which Congress does not readily interfere.

In a recent line of authority, we have acknowledged the unique nature of state decisions that "go to the heart of representative government." Sugarman v. Dougall, 413 U. S. 634, 647 (1973). Sugarman was the first in a series of cases to consider the restrictions imposed by the Equal Protection Clause of the Fourteenth Amendment on the ability of state and local governments to prohibit aliens from public employment. In that case, the Court struck down under the Equal Protection Clause a New York City law that provided a flat ban against the employment of aliens in a wide variety of city jobs. Ibid.

The Court did not hold, however, that alienage could never justify exclusion from public employment. We recognized explicitly the States' constitutional power to establish the qualifications for those who would govern:

"Just as `the Framers of the Constitution intended the States to keep for themselves, as provided in the Tenth [462] Amendment, the power to regulate elections,' Oregon v. Mitchell, 400 U. S. 112, 124-125 (1970) (footnote omitted) (opinion of Black, J.); see id., at 201 (opinion of Harlan, J.), and id., at 293-294 (opinion of STEWART, J.), "[e]ach State has the power to prescribe the qualifications of its officers and the manner in which they shall be chosen." Boyd v. Thayer, 143 U. S. 135, 161 (1892). See Luther v. Borden, 7 How. 1, 41 (1849); Pope v. Williams, 193 U. S. 621, 632-633 (1904). Such power inheres in the State by virtue of its obligation, already noted above, `to preserve the basic conception of a political community.' Dunn v. Blumstein, 405 U. S. [330, 344 (1972)]. And this power and responsibility of the State applies, not only to the qualifications of voters, but also to persons holding state elective and important nonelective executive, legislative, and judicial positions, for officers who participate directly in the formulation, execution, or review of broad public policy perform functions that go to the heart of representative government." Ibid.

We explained that, while the Equal Protection Clause provides a check on such state authority, "our scrutiny will not be so demanding where we deal with matters resting firmly within a State's constitutional prerogatives." Id., at 648. This rule "is no more than . . . a recognition of a State's constitutional responsibility for the establishment and operation of its own government, as well as the qualifications of an appropriately designated class of public office holders. U. S. Const. Art. IV, § 4; U. S. Const. Amdt. X; Luther v. Borden, supra; see In re Duncan, 139 U. S. 449, 461 (1891)." Ibid.

In several subsequent cases we have applied the "political function" exception to laws through which States exclude aliens from positions "intimately related to the process of democratic self-government." See Bernal v. Fainter, 467 U. S. 216, 220 (1984). See also Nyquist v. Mauclet, 432 U. S. 1, 11 (1977); Foley v. Connelie, 435 U. S. 291, 295-296 [463] (1978); Ambach v. Norwick, 441 U. S. 68, 73-74 (1979); Cabell v. Chavez-Salido, 454 U. S. 432, 439-441 (1982). "We have . . . lowered our standard of review when evaluating the validity of exclusions that entrust only to citizens important elective and nonelective positions whose operations `go to the heart of representative government.'" Bernal, 467 U. S., at 221 (citations omitted).

These cases stand in recognition of the authority of the people of the States to determine the qualifications of their most important government officials.^[2] It is an authority that lies at "`the heart of representative government.'" Ibid. It is a power reserved to the States under the Tenth Amendment and guaranteed them by that provision of the Constitution under which the United States "guarantee[s] to every State in this Union a Republican Form of Government." U. S. Const., Art. IV, § 4. See Sugarman, supra, at 648 (citing the Guarantee Clause and the Tenth Amendment). See also Merritt, 88 Colum. L. Rev., at 50-55.

The authority of the people of the States to determine the qualifications of their government officials is, of course, not without limit. Other constitutional provisions, most notably the Fourteenth Amendment, proscribe certain qualifications; our review of citizenship requirements under the political function exception is less exacting, but it is not absent. [464] Here, we must decide what Congress did in extending the ADEA to the States, pursuant to its powers under the Commerce Clause. See EEOC v. Wyoming, 460 U. S. 226 (1983) (the extension of the ADEA to employment by state and local governments was a valid exercise of Congress' powers under the Commerce Clause). As against Congress' powers "[t]o regulate Commerce . . . among the several States," U. S. Const., Art. I, § 8, cl. 3, the authority of the people of the States to determine the qualifications of their government officials may be inviolate.

We are constrained in our ability to consider the limits that the state-federal balance places on Congress' powers under the Commerce Clause. See Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S. 528 (1985) (declining to review limitations placed on Congress' Commerce Clause powers by our federal system). But there is no need to do so if we hold that the ADEA does not apply to state judges. Application of the plain statement rule thus may avoid a potential constitutional problem. Indeed, inasmuch as this Court in Garcia has left primarily to the political process the protection of the States against intrusive exercises of Congress' Commerce Clause powers, we must be absolutely certain that Congress intended such an exercise. "[T]o give the state-displacing weight of federal law to mere congressional ambiguity would evade the very procedure for lawmaking on which Garcia relied to protect states' interests." L. Tribe, American Constitutional Law § 6-25, p. 480 (2d ed. 1988).

In 1974, Congress extended the substantive provisions of the ADEA to include the States as employers. Pub. L. 93-259, § 28(a), 88 Stat. 74, 29 U. S. C. § 630(b)(2). At the same time, Congress amended the definition of "employee" to exclude all elected and most high-ranking government officials. Under the Act, as amended:

[465] "The term `employee' means an individual employed by any employer except that the term `employee' shall not include any person elected to public office in any State or political subdivision of any State by the qualified voters thereof, or any person chosen by such officer to be on such officer's personal staff, or an appointee on the policymaking level or an immediate adviser with respect to the exercise of the constitutional or legal powers of the office." 29 U. S. C. § 630(f).

Governor Ashcroft contends that the § 630(f) exclusion of certain public officials also excludes judges, like petitioners, who are appointed to office by the Governor and are then subject to retention election. The Governor points to two passages in § 630(f). First, he argues, these judges are selected by an elected official and, because they make policy, are "appointee[s] on the policymaking level."

Petitioners counter that judges merely resolve factual disputes and decide questions of law; they do not make policy. Moreover, petitioners point out that the policymaking-level exception is part of a trilogy, tied closely to the electedofficial exception. Thus, the Act excepts elected officials and: (1) "any person chosen by such officer to be on such officer's personal staff"; (2) "an appointee on the policymaking level"; and (3) "an immediate advisor with respect to the exercise of the constitutional or legal powers of the office." Applying the maxim of statutory construction noscitur a sociis — that a word is known by the company it keeps — petitioners argue that since (1) and (3) refer only to those in close working relationships with elected officials, so too must (2). Even if it can be said that judges may make policy, petitioners contend, they do not do so at the behest of an elected official.

Governor Ashcroft relies on the plain language of the statute: It exempts persons appointed "at the policymaking level." The Governor argues that state judges, in fashioning and applying the common law, make policy. Missouri is a [466] common law state. See Mo. Rev. Stat. § 1.010 (1986) (adopting "[t]he common law of England" consistent with federal and state law). The common law, unlike a constitution or statute, provides no definitive text; it is to be derived from the interstices of prior opinions and a well-considered judgment of what is best for the community. As Justice Holmes put it:

"The very considerations which judges most rarely mention, and always with an apology, are the secret root from which the law draws all the juices of life. I mean, of course, considerations of what is expedient for the community concerned. Every important principle which is developed by litigation is in fact and at bottom the result of more or less definitely understood views of public policy; most generally, to be sure, under our practice and traditions, the unconscious result of instinctive preferences and inarticulate convictions, but nonetheless traceable to views of public policy in the last analysis." O. Holmes, The Common Law 35-36 (1881).

Governor Ashcroft contends that Missouri judges make policy in other ways as well. The Missouri Supreme Court and Courts of Appeals have supervisory authority over inferior courts. Mo. Const., Art. V, §4. The Missouri Supreme Court has the constitutional duty to establish rules of practice and procedure for the Missouri court system, and inferior courts exercise policy judgment in establishing local rules of practice. See Mo. Const., Art. V, § 5. The state courts have supervisory powers over the state bar, with the Missouri Supreme Court given the authority to develop disciplinary rules. See Mo. Rev. Stat. §§484.040, 484.200-484.270 (1986); Rules Governing the Missouri Bar and the Judiciary (1991).

The Governor stresses judges' policymaking responsibilities, but it is far from plain that the statutory exception requires that judges actually make policy. The statute refers to appointees "on the policymaking level," not to appointees "who make policy." It may be sufficient that the appointee [467] is in a position requiring the exercise of discretion concerning issues of public importance. This certainly describes the bench, regardless of whether judges might be considered policymakers in the same sense as the executive or legislature.

Nonetheless, "appointee at the policymaking level," particularly in the context of the other exceptions that surround it, is an odd way for Congress to exclude judges; a plain statement that judges are not "employees" would seem the most efficient phrasing. But in this case we are not looking for a plain statement that judges are excluded. We will not read the ADEA to cover state judges unless Congress has made it clear that judges are included. This does not mean that the Act must mention judges explicitly, though it does not. Cf. Dellmuth v. Muth, 491 U. S. 223, 233 (1989) (SCALIA, J., concurring). Rather, it must be plain to anyone reading the Act that it covers judges. In the context of a statute that plainly excludes most important state public officials, "appointee on the policymaking level" is sufficiently broad that we cannot conclude that the statute plainly covers appointed state judges. Therefore, it does not.

The ADEA plainly covers all state employees except those excluded by one of the exceptions. Where it is unambiguous that an employee does not fall within one of the exceptions, the Act states plainly and unequivocally that the employee is included. It is at least ambiguous whether a state judge is an "appointee on the policymaking level."

Governor Ashcroft points also to the "person elected to public office" exception. He contends that because petitioners — although appointed to office initially — are subject to retention election, they are "elected to public office" under the ADEA. Because we conclude that petitioners fall presumptively under the policymaking-level exception, we need not answer this question.

The extension of the ADEA to employment by state and local governments was a valid exercise of Congress' powers [468] under the Commerce Clause. EEOC v. Wyoming, 460 U. S. 226 (1983). In Wyoming, we reserved the questions whether Congress might also have passed the ADEA extension pursuant to its powers under § 5 of the Fourteenth Amendment, and whether the extension would have been a valid exercise of that power. Id., at 243, and n. 18. We noted, however, that the principles of federalism that constrain Congress' exercise of its Commerce Clause powers are attenuated when Congress acts pursuant to its powers to enforce the Civil War Amendments. Id., at 243, and n. 18, citing City of Rome v. United States, 446 U. S. 156, 179 (1980). This is because those "Amendments were specifically designed as an expansion of federal power and an intrusion on state sovereignty." Id., at 179. One might argue, therefore, that if Congress passed the ADEA extension under its § 5 powers, the concerns about federal intrusion into state government that compel the result in this case might carry less weight.

By its terms, the Fourteenth Amendment contemplates interference with state authority: "No State shall . . . deny to any person within its jurisdiction the equal protection of the laws." U. S. Const., Amdt. 14. But this Court has never held that the Amendment may be applied in complete disregard for a State's constitutional powers. Rather, the Court has recognized that the States' power to define the qualifications of their officeholders has force even as against the proscriptions of the Fourteenth Amendment.

We return to the political-function cases. In Sugarman, the Court noted that "aliens as a class `are a prime example of a "discrete and insular" minority (see United States v. Carolene Products Co., 304 U. S. 144, 152-153, n. 4 (1938)),' and that classifications based on alienage are `subject to close judicial scrutiny.'" 413 U. S., at 642, quoting Graham v. Richardson, 403 U. S. 365, 372 (1971). The Sugarman Court held that New York City had insufficient interest in preventing aliens from holding a broad category of public [469] jobs to justify the blanket prohibition. 413 U. S., at 647. At the same time, the Court established the rule that scrutiny under the Equal Protection Clause "will not be so demanding where we deal with matters resting firmly within a State's constitutional prerogatives." Id., at 648. Later cases have reaffirmed this practice. See Foley v. Connelie, 435 U. S. 291 (1978); Ambach v. Norwick, 441 U. S. 68 (1979); Cabell v. Chavez-Salido, 454 U. S. 432 (1982). These cases demonstrate that the Fourteenth Amendment does not override all principles of federalism.

Of particular relevance here is Pennhurst State School and Hospital v. Halderman, 451 U. S. 1 (1981). The question in that case was whether Congress, in passing a section of the Developmentally Disabled Assistance and Bill of Rights Act, 42 U. S. C. § 6010 (1982 ed.), intended to place an obligation on the States to provide certain kinds of treatment to the disabled. Respondent Halderman argued that Congress passed § 6010 pursuant to § 5 of the Fourteenth Amendment, and therefore that it was mandatory on the States, regardless of whether they received federal funds. Petitioner and the United States, as respondent, argued that, in passing § 6010, Congress acted pursuant to its spending power alone. Consequently, § 6010 applied only to States accepting federal funds under the Act.

The Court was required to consider the "appropriate test for determining when Congress intends to enforce" the guarantees of the Fourteenth Amendment. 451 U. S., at 16. We adopted a rule fully cognizant of the traditional power of the States: "Because such legislation imposes congressional policy on a State involuntarily, and because it often intrudes on traditional state authority, we should not quickly attribute to Congress an unstated intent to act under its authority to enforce the Fourteenth Amendment." Ibid. Because Congress nowhere stated its intent to impose mandatory obligations on the States under its § 5 powers, we concluded that Congress did not do so. Ibid.

[470] The Pennhurst rule looks much like the plain statement rule we apply today. In EEOC v. Wyoming, the Court explained that Pennhurst established a rule of statutory construction to be applied where statutory intent is ambiguous. 460 U. S., at 244, n. 18. In light of the ADEA's clear exclusion of most important public officials, it is at least ambiguous whether Congress intended that appointed judges nonetheless be included. In the face of such ambiguity, we will not attribute to Congress an intent to intrude on state governmental functions regardless of whether Congress acted pursuant to its Commerce Clause powers or § 5 of the Fourteenth Amendment.

Petitioners argue that, even if they are not covered by the ADEA, the Missouri Constitution's mandatory retirement provision for judges violates the Equal Protection Clause of the Fourteenth Amendment to the United States Constitution. Petitioners contend that there is no rational basis for the decision of the people of Missouri to preclude those aged 70 and over from serving as their judges. They claim that the mandatory retirement provision makes two irrational distinctions: between judges who have reached age 70 and younger judges, and between judges 70 and over and other state employees of the same age who are not subject to mandatory retirement.

Petitioners are correct to assert their challenge at the level of rational basis. This Court has said repeatedly that age is not a suspect classification under the Equal Protection Clause. See Massachusetts Bd. of Retirement v. Murgia, 427 U. S. 307, 313-314 (1976); Vance v. Bradley, 440 U. S. 93, 97 (1979); Cleburne v. Cleburne Living Center, Inc., 473 U. S. 432, 441 (1985). Nor do petitioners claim that they have a fundamental interest in serving as judges. The State need therefore assert only a rational basis for its age classification. See Murgia, supra, at 314; Bradley, 440 U. S., at 97. In cases where a classification burdens neither a suspect [471] group nor a fundamental interest, "courts are quite reluctant to overturn governmental action on the ground that it denies equal protection of the laws." Ibid. In this case, we are dealing not merely with government action, but with a state constitutional provision approved by the people of Missouri as a whole. This constitutional provision reflects both the considered judgment of the state legislature that proposed it and that of the citizens of Missouri who voted for it. See 1976 Mo. Laws 812 (proposing the mandatory retirement provision of § 26); Mo. Const., Art. XII, §§ 2(a), 2(b) (describing the amendment process). "[W]e will not overturn such a [law] unless the varying treatment of different groups or persons is so unrelated to the achievement of any combination of legitimate purposes that we can only conclude that the [people's] actions were irrational." Bradley, supra, at 97. See also Pennell v. San Jose, 485 U. S. 1, 14 (1988).

Governor Ashcroft cites O'Neil v. Baine, 568 S. W. 2d 761 (Mo. 1978) (en banc), as a fruitful source of rational bases. In O'Neil, the Missouri Supreme Court — to whom Missouri Constitution Article V, § 26, applies — considered an equal protection challenge to a state statute that established a mandatory retirement age of 70 for state magistrate and probate judges. The court upheld the statute, declaring numerous legitimate state objectives it served: "The statute draws a line at a certain age which attempts to uphold the high competency for judicial posts and which fulfills a societal demand for the highest caliber of judges in the system"; "the statute... draws a legitimate line to avoid the tedious and often perplexing decisions to determine which judges after a certain age are physically and mentally qualified and those who are not"; "mandatory retirement increases the opportunity for qualified persons... to share in the judiciary and permits an orderly attrition through retirement"; "such a mandatory provision also assures predictability and ease in establishing and administering judges' pension plans." Id., at 766-767. Any one of these explanations is sufficient to rebut the claim [472] that "the varying treatment of different groups or persons [in § 26] is so unrelated to the achievement of any combination of legitimate purposes that we can only conclude that the [people's] actions were irrational." Bradley, supra, at 97.

The people of Missouri have a legitimate, indeed compelling, interest in maintaining a judiciary fully capable of performing the demanding tasks that judges must perform. It is an unfortunate fact of life that physical and mental capacity sometimes diminish with age. See Bradley, supra, at 111-112; Murgia, supra, at 315. The people may therefore wish to replace some older judges. Voluntary retirement will not always be sufficient. Nor may impeachment — with its public humiliation and elaborate procedural machinery — serve acceptably the goal of a fully functioning judiciary. See Mo. Const., Art. VII, §§ 1-3.

The election process may also be inadequate. Whereas the electorate would be expected to discover if their governor or state legislator were not performing adequately and vote the official out of office, the same may not be true of judges. Most voters never observe state judges in action, nor read judicial opinions. State judges also serve longer terms of office than other public officials, making them — deliberately — less dependent on the will of the people. Compare Mo. Const., Art. V, § 19 (Supreme Court justices and Court of Appeals judges serve 12-year terms; Circuit Court judges 6 years), with Mo. Const., Art. IV, § 17 (Governor, Lieutenant Governor, secretary of state, state treasurer, and attorney general serve 4-year terms) and Mo. Const., Art. III, § 11 (state representatives serve 2-year terms; state senators 4 years). Most of these judges do not run in ordinary elections. See Mo. Const., Art. V, § 25(a). The people of Missouri rationally could conclude that retention elections — in which state judges run unopposed at relatively long intervals — do not serve as an adequate check on judges whose performance is deficient. Mandatory retirement is a reasonable response to this dilemma.

[473] This is also a rational explanation for the fact that state judges are subject to a mandatory retirement provision, while other state officials —whose performance is subject to greater public scrutiny, and who are subject to more standard elections — are not. Judges' general lack of accountability explains also the distinction between judges and other state employees, in whom a deterioration in performance is more readily discernible and who are more easily removed.

The Missouri mandatory retirement provision, like all legal classifications, is founded on a generalization. It is far from true that all judges suffer significant deterioration in performance at age 70. It is probably not true that most do. It may not be true at all. But a State "`does not violate the Equal Protection Clause merely because the classifications made by its laws are imperfect.'" Murgia, 427 U. S., at 316, quoting Dandridge v. Williams, 397 U. S. 471, 485 (1970). "In an equal protection case of this type ... those challenging the . . . judgment [of the people] must convince the court that the ... facts on which the classification is apparently based could not reasonably be conceived to be true by the ... decisionmaker." Bradley, 440 U. S., at 111. The people of Missouri rationally could conclude that the threat of deterioration at age 70 is sufficiently great, and the alternatives for removal sufficiently inadequate, that they will require all judges to step aside at age 70. This classification does not violate the Equal Protection Clause.

The people of Missouri have established a qualification for those who would be their judges. It is their prerogative as citizens of a sovereign State to do so. Neither the ADEA nor the Equal Protection Clause prohibits the choice they have made. Accordingly, the judgment of the Court of Appeals is

Affirmed.

I agree with the majority that neither the Age Discrimination in Employment Act of 1967 (ADEA) nor the Equal Protection Clause prohibits Missouri's mandatory retirement provision as applied to petitioners, and I therefore concur in the judgment and in Parts I and III of the majority's opinion. I cannot agree, however, with the majority's reasoning in Part II of its opinion, which ignores several areas of well-established precedent and announces a rule that is likely to prove both unwise and infeasible. That the majority's analysis in Part II is completely unnecessary to the proper resolution of this case makes it all the more remarkable.

In addition to petitioners' equal protection claim, we granted certiorari to decide the following question:

"Whether appointed Missouri state court judges are `appointee[s] on the policymaking level' within the meaning of the Age Discrimination in Employment Act (`ADEA'), 28 U. S. C. §§ 621-34 (1982 & Supp. V 1987), and therefore exempted from the ADEA's general prohibition of mandatory retirement and thus subject to the mandatory retirement provision of Article V, Section 26 of the Missouri Constitution." Pet. for Cert. i.

The majority, however, chooses not to resolve that issue of statutory construction. Instead, it holds that whether or not the ADEA can fairly be read to exclude state judges from its scope, "[w]e will not read the ADEA to cover state judges unless Congress has made it clear that judges are included." Ante, at 467 (emphasis in original). I cannot agree with this "plain statement" rule because it is unsupported by the decisions upon which the majority relies, contrary to our Tenth Amendment jurisprudence, and fundamentally unsound.

[475] Among other things, the ADEA makes it "unlawful for an employer— (1) to fail or refuse to hire or to discharge any individual or otherwise discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual's age." 29 U. S. C. § 623(a). In 1974, Congress amended the definition of "employer" in the ADEA to include "a State or political subdivision of a State." § 630(b)(2). With that amendment, "there is no doubt what the intent of Congress was: to extend the application of the ADEA to the States." EEOC v. Wyoming, 460 U. S. 226, 244, n. 18 (1983).

The dispute in this case therefore is not whether Congress has outlawed age discrimination by the States. It clearly has. The only question is whether petitioners fall within the definition of "employee" in the Act, § 630(f), which contains exceptions for elected officials and certain appointed officials. If petitioners are "employee[s]," Missouri's mandatory retirement provision clearly conflicts with the antidiscrimination provisions of the ADEA. Indeed, we have noted that the "policies and substantive provisions of the [ADEA] apply with especial force in the case of mandatory retirement provisions." Western Air Lines, Inc. v. Criswell, 472 U. S. 400, 410 (1985). Pre-emption therefore is automatic, since "state law is pre-empted to the extent that it actually conflicts with federal law." Pacific Gas & Elec. Co. v. State Energy Resources Conservation and Development Comm'n, 461 U. S. 190, 204 (1983). The majority's federalism concerns are irrelevant to such "actual conflict" pre-emption. "`The relative importance to the State of its own law is not material when there is a conflict with a valid federal law, for the Framers of our Constitution provided that the federal law must prevail.'" Fidelity Federal Say. & Loan Assn. v. De la Cuesta, 458 U. S. 141, 153 (1982), quoting Free v. Bland, 369 U. S. 663, 666 (1962).

While acknowledging this principle of federal legislative supremacy, see ante, at 460, the majority nevertheless imposes [476] upon Congress a "plain statement" requirement. The majority claims to derive this requirement from the plain statement approach developed in our Eleventh Amendment cases, see, e. g., Atascadero State Hospital v. Scanlon, 473 U. S. 234, 243 (1985), and applied two Terms ago in Will v. Michigan Dept. of State Police, 491 U. S. 58, 65 (1989). The issue in those cases, however, was whether Congress intended a particular statute to extend to the States at all. In Atascadero, for example, the issue was whether States could be sued under § 504 of the Rehabilitation Act of 1973, 29 U. S. C. § 794. Similarly, the issue in Will was whether States could be sued under 42 U. S. C. § 1983. In the present case, by contrast, Congress has expressly extended the coverage of the ADEA to the States and their employees. Its intention to regulate age discrimination by States is thus "unmistakably clear in the language of the statute." Atascadero, supra, at 242. See Davidson v. Board of Governors of State Colleges and Universities, 920 F. 2d 441, 443 (CA7 1990) (ADEA satisfies "clear statement" requirement). The only dispute is over the precise details of the statute's application. We have never extended the plain statement approach that far, and the majority offers no compelling reason for doing so.

The majority also relies heavily on our cases addressing the constitutionality of state exclusion of aliens from public employment. See ante, at 461-463, 468-470. In those cases, we held that although restrictions based on alienage ordinarily are subject to strict scrutiny under the Equal Protection Clause, see Graham v. Richardson, 403 U. S. 365, 372 (1971), the scrutiny will be less demanding for exclusion of aliens "from positions intimately related to the process of democratic self-government." Bernal v. Fainter, 467 U. S. 216, 220 (1984). This narrow "political-function" exception to the strict-scrutiny standard is based on the "State's historical power to exclude aliens from participation in its [477] democratic political institutions." Sugarman v. Dougall, 413 U. S. 634, 648 (1973).

It is difficult to see how the "political-function" exception supports the majority's plain statement rule. First, the exception merely reflects a determination of the scope of the rights of aliens under the Equal Protection Clause. Reduced scrutiny is appropriate for certain political functions because "the right to govern is reserved to citizens." Foley v. Connelie, 435 U. S. 291, 297 (1978); see also Sugarman, supra, at 648-649. This conclusion in no way establishes a method for interpreting rights that are statutorily created by Congress, such as the protection from age discrimination in the ADEA. Second, it is one thing to limit judicially created scrutiny, and it is quite another to fashion a restraint on Congress' legislative authority, as does the majority; the latter is both counter-majoritarian and an intrusion on a coequal branch of the Federal Government. Finally, the majority does not explicitly restrict its rule to "functions that go to the heart of representative government," 413 U. S., at 647, and may in fact be extending it much further to all "state governmental functions." See ante, at 470.

The majority's plain statement rule is not only unprecedented, it directly contravenes our decisions in Garcia v. San Antonio Metropolitan Transit Authority, 469 U. S. 528 (1985), and South Carolina v. Baker, 485 U. S. 505 (1988). In those cases we made it clear "that States must find their protection from congressional regulation through the national political process, not through judicially defined spheres of unregulable state activity." Id., at 512. We also rejected as "unsound in principle and unworkable in practice" any test for state immunity that requires a judicial determination of which state activities are "`traditional,'" "`integral,'" or "`necessary.'" Garcia, supra, at 546. The majority disregards those decisions in its attempt to carve out areas of state activity that will receive special protection from federal legislation.

[478] The majority's approach is also unsound because it will serve only to confuse the law. First, the majority fails to explain the scope of its rule. Is the rule limited to federal regulation of the qualifications of state officials? See ante, at 464. Or does it apply more broadly to the regulation of any "state governmental functions"? See ante, at 470. Second, the majority does not explain its requirement that Congress' intent to regulate a particular state activity be "plain to anyone reading [the federal statute]." See ante, at 467. Does that mean that it is now improper to look to the purpose or history of a federal statute in determining the scope of the statute's limitations on state activities? If so, the majority's rule is completely inconsistent with our pre-emption jurisprudence. See, e. g., Hillsborough County v. Automated Medical Laboratories, Inc., 471 U. S. 707, 715 (1985) (pre-emption will be found where there is a "`clear and manifest purpose'" to displace state law) (emphasis added). The vagueness of the majority's rule undoubtedly will lead States to assert that various federal statutes no longer apply to a wide variety of state activities if Congress has not expressly referred to those activities in the statute. Congress, in turn, will be forced to draft long and detailed lists of which particular state functions it meant to regulate.

The imposition of such a burden on Congress is particularly out of place in the context of the ADEA. Congress already has stated that all "individual[s] employed by any employer" are protected by the ADEA unless they are expressly excluded by one of the exceptions in the definition of "employee." See 29 U. S. C. § 630(f). The majority, however, turns the statute on its head, holding that state judges are not protected by the ADEA because "Congress has [not] made it clear that judges are included." Ante, at 467 (emphasis in original). Cf. EEOC v. Wyoming, 460 U. S. 226 (1983), where we held that state game wardens are covered by the ADEA, even though such employees are not expressly included within the ADEA's scope.

[479] The majority asserts that its plain statement rule is helpful in avoiding a "potential constitutional problem." Ante, at 464. It is far from clear, however, why there would be a constitutional problem if the ADEA applied to state judges, in light of our decisions in Garcia and Baker, discussed above. As long as "the national political process did not operate in a defective manner, the Tenth Amendment is not implicated." Baker, supra, at 513. There is no claim in this case that the political process by which the ADEA was extended to state employees was inadequate to protect the States from being "unduly burden[ed]" by the Federal Government. See Garcia, supra, at 556. In any event, as discussed below, a straightforward analysis of the ADEA's definition of "employee" reveals that the ADEA does not apply here. Thus, even if there were potential constitutional problems in extending the ADEA to state judges, the majority's proposed plain statement rule would not be necessary to avoid them in this case. Indeed, because this case can be decided purely on the basis of statutory interpretation, the majority's announcement of its plain statement rule, which purportedly is derived from constitutional principles, violates our general practice of avoiding the unnecessary resolution of constitutional issues.

My disagreement with the majority does not end with its unwarranted announcement of the plain statement rule. Even more disturbing is its treatment of Congress' power under § 5 of the Fourteenth Amendment. See ante, at 467-470. Section 5 provides that "[t]he Congress shall have power to enforce, by appropriate legislation, the provisions of this article." Despite that sweeping constitutional delegation of authority to Congress, the majority holds that its plain statement rule will apply with full force to legislation enacted to enforce the Fourteenth Amendment. The majority states: "In the face of . . . ambiguity, we will not attribute to Congress an intent to intrude on state governmental functions regardless of whether Congress acted pursuant to its [480] Commerce Clause powers or § 5 of the Fourteenth Amendment." Ante, at 470 (emphasis added).^[3]

The majority's failure to recognize the special status of legislation enacted pursuant to § 5 ignores that, unlike Congress' Commerce Clause power, "[w]hen Congress acts pursuant to § 5, not only is it exercising legislative authority that is plenary within the terms of the constitutional grant, it is exercising that authority under one section of a constitutional Amendment whose other sections by their own terms embody limitations on state authority." Fitzpatrick v. Bitzer, 427 U. S. 445, 456 (1976). Indeed, we have held that "principles of federalism that might otherwise be an obstacle to congressional authority are necessarily overridden by the power to enforce the Civil War Amendments `by appropriate legislation.' Those Amendments were specifically designed as an expansion of federal power and an intrusion on state sovereignty." City of Rome v. United States, 446 U. S. 156, 179 (1980); see also EEOC v. Wyoming, supra, at 243, n. 18.

The majority relies upon Pennhurst State School and Hospital v. Halderman, 451 U. S. 1 (1981), see ante, at 469-470, but that case does not support its approach. There, the Court merely stated that "we should not quickly attribute to Congress an unstated intent to act under its authority to enforce the Fourteenth Amendment." 451 U. S., at 16. In other words, the Pennhurst presumption was designed only to answer the question whether a particular piece of legislation [481] was enacted pursuant to § 5. That is very different from the majority's apparent holding that even when Congress is acting pursuant to § 5, it nevertheless must specify the precise details of its enactment.

The majority's departures from established precedent are even more disturbing when it is realized, as discussed below, that this case can be affirmed based on simple statutory construction.

The statute at issue in this case is the ADEA's definition of "employee," which provides:

"The term `employee' means an individual employed by any employer except that the term `employee' shall not include any person elected to public office in any State or political subdivision of any State by the qualified voters thereof, or any person chosen by such officer to be on such officer's personal staff, or an appointee on the policymaking level or an immediate adviser with respect to the exercise of the constitutional or legal powers of the office. The exemption set forth in the preceding sentence shall not include employees subject to the civil service laws of a State government, governmental agency, or political subdivision." 29 U. S. C. § 630(f).

A parsing of that definition reveals that it excludes from the definition of "employee" (and thus the coverage of the ADEA) four types of (noncivil service) state and local employees: (1) persons elected to public office; (2) the personal staff of elected officials; (3) persons appointed by elected officials to be on the policymaking level; and (4) the immediate advisers of elected officials with respect to the constitutional or legal powers of the officials' offices.

The question before us is whether petitioners fall within the third exception. Like the Court of Appeals, see 898 F. 2d 598, 600 (CA8 1990), I assume that petitioners, who were initially appointed to their positions by the Governor of [482] Missouri, are "appointed" rather than "elected" within the meaning of the ADEA. For the reasons below, I also conclude that petitioners are "on the policymaking level."^[4]

"Policy" is defined as "a definite course or method of action selected (as by a government, institution, group, or individual) from among alternatives and in the light of given conditions to guide and usu[ally] determine present and future decisions." Webster's Third New International Dictionary 1754 (1976). Applying that definition, it is clear that the decisionmaking engaged in by common-law judges, such as petitioners, places them "on the policymaking level." In resolving disputes, although judges do not operate with unconstrained discretion, they do choose "from among alternatives" and elaborate their choices in order "to guide and ... determine present and future decisions." The quotation from Justice Holmes in the majority's opinion, see ante, at 466, is an eloquent description of the policymaking nature of the judicial function. Justice Cardozo also stated it well:

"Each [common-law judge] indeed is legislating within the limits of his competence. No doubt the limits for the judge are narrower. He legislates only between gaps. He fills the open spaces in the law. . . . [W]ithin the confines of these open spaces and those of precedent and tradition, choice moves with a freedom which stamps its action as creative. The law which is the resulting product is not found, but made." B. Cardozo, The Nature of the Judicial Process 113-115 (1921).

[483] Moreover, it should be remembered that the statutory exception refers to appointees "on the policymaking level," not "policymaking employees." Thus, whether or not judges actually make policy, they certainly are on the same level as policymaking officials in other branches of government and therefore are covered by the exception. The degree of responsibility vested in judges, for example, is comparable to that of other officials that have been found by the lower courts to be on the policymaking level. See, e. g., EEOC v. Reno, 758 F. 2d 581 (CA11 1985) (assistant state attorney); EEOC v. Board of Trustees of Wayne Cty. Community College, 723 F. 2d 509 (CA6 1983) (president of community college).

Petitioners argue that the "appointee[s] on the policymaking level" exception should be construed to apply "only to persons who advise or work closely with the elected official that chose the appointee." Brief for Petitioners 18. In support of that claim, petitioners point out that the exception is "sandwiched" between the "personal staff" and "immediate adviser" exceptions in § 630(f), and thus should be read as covering only similar employees.

Petitioners' premise, however, does not prove their conclusion. It is true that the placement of the "appointee" exception between the "personal staff" and "immediate adviser" exceptions suggests a similarity among the three. But the most obvious similarity is simply that each of the three sets of employees are connected in some way with elected officials: The first and third sets have a certain working relationship with elected officials, while the second is appointed by elected officials. There is no textual support for concluding that the second set must also have a close working relationship with elected officials. Indeed, such a reading would tend to make the "appointee" exception superfluous since the "personal staff" and "immediate adviser" exceptions would seem to cover most appointees who are in a close working relationship with elected officials.

[484] Petitioners seek to rely on legislative history, but it does not help their position. There is little legislative history discussing the definition of "employee" in the ADEA, so petitioners point to the legislative history of the identical definition in Title VII of the Civil Rights Act of 1964, 42 U. S. C. § 2000e(f). If anything, that history tends to confirm that the "appointee[s] on the policymaking level" exception was designed to exclude from the coverage of the ADEA all high-level appointments throughout state government structures, including judicial appointments.

For example, during the debates concerning the proposed extension of Title VII to the States, Senator Ervin repeatedly expressed his concern that the (unamended) definition of "employee" would be construed to reach those "persons who exercise the legislative, executive, and judicial powers of the States and political subdivisions of the States." 118 Cong. Rec. 1838 (1972) (emphasis added). Indeed, he expressly complained that "[t]here is not even an exception in the [unamended] bill to the effect that the EEOC will not have jurisdiction over ... State judges, whether they are elected or appointed to office." Id., at 1677. Also relevant is Senator Taft's comment that, in order to respond to Senator Ervin's concerns, he was willing to agree to an exception not only for elected officials, but also for "those at the top decisionmaking levels in the executive and judicial branch as well." Id., at 1838.

The definition of "employee" subsequently was modified to exclude the four categories of employees discussed above. The Conference Committee that added the "appointee[s] on the policymaking level" exception made clear the separate nature of that exception:

"It is the intention of the conferees to exempt elected officials and members of their personal staffs, and persons appointed by such elected officials as advisors or to policymaking positions at the highest levels of the departments or agencies of State or local governments, such as [485] cabinet officers, and persons with comparable responsibilities at the local level." H. R. Conf. Rep. No. 92-899, pp. 15-16 (1972) (emphasis added).

The italicized "or" in that statement indicates, contrary to petitioners' argument, that appointed officials need not be advisers to be covered by the exception. Rather, it appears that "Congress intended two categories: policymakers, who need not be advisers; and advisers, who need not be policymakers." EEOC v. Massachusetts, 858 F. 2d 52, 56 (CA1 1988). This reading is confirmed by a statement by one of the House Managers, Representative Erlenborn, who explained that "[i]n the conference, an additional qualification was added, exempting those people appointed by officials at the State and local level in policymaking positions." 118 Cong. Rec., at 7567.

In addition, the phrase "the highest levels" in the Conference Report suggests that Congress' intent was to limit the exception "down the chain of command, and not so much across agencies or departments." EEOC v. Massachusetts, 858 F. 2d, at 56. I also agree with the First Circuit's conclusion that even lower court judges fall within the exception because "each judge, as a separate and independent judicial officer, is at the very top of his particular `policymaking' chain of command, responding . . . only to a higher appellate court." Ibid.

For these reasons, I would hold that petitioners are excluded from the coverage of the ADEA because they are "appointee[s] on the policymaking level" under 29 U. S. C. § 630(f).^[5]

[486] I join Parts I and III of the Court's opinion and concur in its judgment.

I agree entirely with the cogent analysis contained in Part I of JUSTICE WHITE'S opinion, ante, at 474-481. For the reasons well stated by JUSTICE WHITE, the question we must resolve is whether appointed Missouri state judges are excluded from the general prohibition of mandatory retirement that Congress established in the federal Age Discrimination in Employment Act of 1967 (ADEA), 29 U. S. C. §§ 621-634. I part company with JUSTICE WHITE, however, in his determination that appointed state judges fall within the narrow exclusion from ADEA coverage that Congress created for an "appointee on the policymaking level." § 630(f).

For two reasons, I do not accept the notion that an appointed state judge is an "appointee on the policymaking level." First, even assuming that judges may be described as policymakers in certain circumstances, the structure and legislative history of the policymaker exclusion make clear that judges are not the kind of policymakers whom Congress intended to exclude from the ADEA's broad reach. Second, [487] whether or not a plausible argument may be made for judges' being policymakers, I would defer to the EEOC's reasonable construction of the ADEA as covering appointed state judges.

Although it may be possible to define an appointed judge as a "policymaker" with only a dictionary as a guide,^[6] we have an obligation to construe the exclusion of an "appointee on the policymaking level" with a sensitivity to the context in which Congress placed it. In construing an undefined statutory term, this Court has adhered steadfastly to the rule that "`"`words grouped in a list should be given related meaning,'"'" Dole v. Steelworkers, 494 U. S. 26, 36 (1990), quoting Massachusetts v. Morash, 490 U. S. 107, 114-115 (1989), quoting Schreiber v. Burlington Northern, Inc., 472 U. S. 1, 8 (1985), quoting Securities Industry Assn. v. Board of Governors, FRS, 468 U. S. 207, 218 (1984), and that "`in expounding a statute, we [are] not . . . guided by a single sentence or member of a sentence, but look to the provisions of [488] the whole law, and to its object and policy.'" Morash, 490 U. S., at 115, quoting Pilot Life Ins. Co. v. Dedeaux, 481 U. S. 41, 51 (1987). Applying these maxims of statutory construction, I conclude that an appointed state judge is not the kind of "policymaker" whom Congress intended to exclude from the protection of the ADEA.

The policymaker exclusion is placed between the exclusion of "any person chosen by such [elected] officer to be on such officer's personal staff" and the exclusion of "an immediate adviser with respect to the exercise of the constitutional or legal powers of the office." See 29 U. S. C. § 630(f). Reading the policymaker exclusion in light of the other categories of employees listed with it, I conclude that the class of "appointee[s] on the policymaking level" should be limited to those officials who share the characteristics of personal staff members and immediate advisers, i. e., those who work closely with the appointing official and are directly accountable to that official. Additionally, I agree with the reasoning of the Second Circuit in EEOC v. Vermont, 904 F. 2d 794 (1990):

"Had Congress intended to except a wide-ranging category of policymaking individuals operating wholly independently of the elected official, it would probably have placed that expansive category at the end of the series, not in the middle." Id., at 798.

Because appointed judges are not accountable to the official who appoints them and are precluded from working closely with that official once they have been appointed, they are not "appointee[s] on the policymaking level" for purposes of 29 U. S. C. § 630(f).^[7]

The evidence of Congress' intent in enacting the policymaking exclusion supports this narrow reading. As noted by JUSTICE WHITE, ante, at 484, there is little in the legislative history of § 630(f) itself to aid our interpretive endeavor. Because Title VII of the Civil Rights Act of 1964, § 701(f), as amended, 42 U. S. C. § 2000e(f), contains language identical to that in the ADEA's policymaking exclusion, however, we accord substantial weight to the legislative history of the cognate Title VII provision in construing § 630(f). See Lorillard v. Pons, 434 U. S. 575, 584 (1978) (noting that "the prohibitions of the ADEA were derived in haec verba from Title VII"). See also Trans World Airlines, Inc. v. Thurston, 469 U. S. 111, 121 (1985); Oscar Mayer & Co. v. Evans, 441 U. S. 750, 756 (1979); EEOC v. Vermont, 904 F. 2d, at 798.

When Congress decided to amend Title VII to include States and local governments as employers, the original bill did not contain any employee exclusion. As JUSTICE WHITE notes, ante, at 484, the absence of a provision excluding certain state employees was a matter of concern for Senator Ervin, who commented that the bill, as reported, did not contain a provision "to the effect that the EEOC will not have jurisdiction over ... State judges, whether they are elected or appointed to office . . . ." 118 Cong. Rec. 1677 (1972). Because this floor comment refers to appointed judges, JUSTICE WHITE concludes that the later amendment containing the exclusion of "an appointee on the policymaking level" was drafted in response to the concerns raised by Senator Ervin and others, ante, at 484-485, and therefore should be read to include judges.

Even if the only legislative history available was the above-quoted statement of Senator Ervin and the final [490] amendment containing the policymaking exclusion, I would be reluctant to accept JUSTICE WHITE'S analysis. It would be odd to conclude that the general exclusion of those "on the policymaking level" was added in response to Senator Ervin's very specific concern about appointed judges. Surely, if Congress had desired to exclude judges — and was responding to a specific complaint that judges would be within the jurisdiction of the EEOC — it would have chosen far clearer language to accomplish this end.^[8] In any case, a more detailed look at the genesis of the policymaking exclusion seriously undermines the suggestion that it was intended to include appointed judges.

After commenting on the absence of an employee exclusion, Senator Ervin proposed the following amendment:

"[T]he term `employee' as set forth in the original act of 1964 and as modified by the pending bill shall not include any person elected to public office in any State or political subdivision of any State by the qualified voters thereof, or any person chosen by such person to advise him in respect to the exercise of the constitutional or legal powers of his office." 118 Cong. Rec. 4483 (1972).

Noticeably absent from this proposed amendment is any reference to those on the policymaking level or to judges. Senator Williams then suggested expanding the proposed amendment to include the personal staff of the elected individual, leading Senators Williams and Ervin to engage in the following discussion about the purpose of the amendment:

[491] "Mr. WILLIAMS: . . . .

"... First, State and local governments are now included under the bill as employers. The amendment would provide, for the purposes of the bill and for the basic law, that an elected individual is not an employee and, th[e]refore, the law could not cover him. The next point is that the elected official would, in his position as an employer, not be covered and would be exempt in the employment of certain individuals.

. . . . .

". . . [B]asically the purpose of the amendment . . . [is] to exempt from coverage those who are chosen by the Governor or the mayor or the county supervisor, whatever the elected official is, and who are in a close personal relationship and an immediate relationship with him. Those who are his first line of advisers. Is that basically the purpose of the Senator's amendment?

"Mr. ERVIN: I would say to my good friend from New Jersey that that is the purpose of the amendment." Id., at 4492-4493.

Following this exchange, Senator Ervin's amendment was expanded to exclude "any person chosen by such officer to be a personal assistant." Id., at 4493. The Senate adopted these amendments, voting to exclude both personal staff members and immediate advisers from the scope of Title VII.

The policymaker exclusion appears to have arisen from Senator Javits' concern that the exclusion for advisers would sweep too broadly, including hundreds of functionaries such as "lawyers, . . . stenographers, subpena servers, researchers, and so forth." Id., at 4097. Senator Javits asked "to have overnight to check into what would be the status of that rather large group of employees," noting that he "realize[d] that . . . Senator [Ervin was] . . . seeking to confine it to the higher officials in a policymaking or policy advising capacity." [492] Ibid. In an effort to clarify his point, Senator Javits later stated:

"The other thing, the immediate advisers, I was thinking more in terms of a cabinet, of a Governor who would call his commissioners a cabinet, or he may have a cabinet composed of three or four executive officials, or five or six, who would do the main and important things. That is what I would define those things expressly to mean." Id., at 4493.

Although Senator Ervin assured Senator Javits that the exclusion of personal staff and advisers affected only the classes of employees that Senator Javits had mentioned, ibid., the Conference Committee eventually adopted a specific exclusion of an "appointee on the policymaking level" as well as the exclusion of personal staff and immediate advisers contained in the Senate bill. In explaining the scope of the exclusion, the conferees stated:

"It is the intention of the conferees to exempt elected officials and members of their personal staffs, and persons appointed by such elected officials as advisors or to policymaking positions at the highest levels of the departments or agencies of State or local governments, such as cabinet officers, and persons with comparable responsibilities at the local level. It is the conferees['] intent that this exemption shall be construed narrowly." S. Conf. Rep. No. 92-681, pp. 15-16 (1972).

The foregoing history decisively refutes the argument that the policymaker exclusion was added in response to Senator Ervin's concern that appointed state judges would be protected by Title VII. Senator Ervin's own proposed amendment did not exclude those on the policymaking level. Indeed, Senator Ervin indicated that all of the policymakers he sought to have excluded from the coverage of Title VII were encompassed in the exclusion of personal staff and immediate advisers. It is obvious that judges are neither staff nor immediate [493] advisers of any elected official. The only indication as to whom Congress understood to be "appointee[s] on the policymaking level" is Senator Javits' reference to members of the Governor's cabinet, echoed in the Conference Committee's use of "cabinet officers" as an example of the type of appointee at the policymaking level excluded from Title VII's definition of "employee." When combined with the Conference Committee's exhortation that the exclusion be construed narrowly, this evidence indicates that Congress did not intend appointed state judges to be excluded from the reach of Title VII or the ADEA.

This Court has held that when a statutory term is ambiguous or undefined, a court construing the statute should defer to a reasonable interpretation of that term proffered by the agency entrusted with administering the statute. See Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 842-843 (1984). Thus, even were I to conclude that one might read the exclusion of an "appointee on the policymaking level" to include state judges, our precedent would compel me to accept the EEOC's contrary reading of the exclusion if it were a "permissible" interpretation of this ambiguous term. Id., at 843. This Court has recognized that "it is axiomatic that the EEOC's interpretation of Title VII, for which it has primary enforcement responsibility, need not be the best one by grammatical or any other standards. Rather, the EEOC's interpretation of ambiguous language need only be reasonable to be entitled to deference." EEOC v. Commercial Office Products Co., 486 U. S. 107, 115 (1988). The EEOC's interpretation of ADEA provisions is entitled to the same deference as its interpretation of analogous provisions in Title VII. See Oscar Mayer & Co. v. Evans, 441 U. S., at 761, citing Griggs v. Duke Power Co., 401 U. S. 424, 434 (1971).

[494] The EEOC consistently has taken the position that an appointed judge is not an "appointee on the policymaking level" within the meaning of 29 U. S. C. § 630(f). See EEOC v. Vermont, 904 F. 2d 794 (CA2 1990); EEOC v. Massachusetts, 858 F. 2d 52 (CA1 1988); EEOC v. Illinois, 721 F. Supp. 156 (ND Ill. 1989). Relying on the legislative history detailed above, the EEOC has asserted that Congress intended the policymaker exclusion to include only "`an elected official's first line advisers.'" EEOC v. Massachusetts, 858 F. 2d, at 55. See also CCH EEOC Decisions (1983) ¶ 6725 (discussing the meaning of the policymaker exclusion under Title VII, and stating that policymakers "must work closely with elected officials and their advisors in developing policies that will implement the overall goals of the elected officials"). As is evident from the foregoing discussion, I believe this to be a correct reading of the statute and its history. At a minimum, it is a "permissible" reading of the indisputably ambiguous term "appointee on the policymaking level." Accordingly, I would defer to the EEOC's reasonable interpretation of this term.^[9]

The Missouri constitutional provision mandating the retirement of a judge who reaches the age of 70 violates the ADEA and is, therefore, invalid.^[10] Congress enacted the ADEA with the express purpose "to promote employment of older persons based on their ability rather than age; to prohibit arbitrary age discrimination in employment; to help employers and workers find ways of meeting problems arising from the impact of age on employment." 29 U. S. C. § 621. Congress provided for only limited exclusions from the coverage of the ADEA, and exhorted courts applying this law to construe such exclusions narrowly. The statute's structure and legislative history reveal that Congress did not intend an appointed state judge to be beyond the scope of the ADEA's protective reach. Further, the EEOC, which is charged with the enforcement of the ADEA, has determined that an appointed state judge is covered by the ADEA. This Court's precedent dictates that we defer to the EEOC's permissible interpretation of the ADEA.

I dissent.

[1] Cathy Ventrell-Monsees filed a brief for the American Association of Retired Persons as amicus curiae urging reversal.

Briefs of amici curiae urging affirmance were filed for the State of Colorado et al. by Scott Harshbarger, Attorney General of Massachusetts, H. Reed Witherby, Special Assistant Attorney General, and Thomas A. Barnico, Assistant Attorney General, and by the Attorneys General for their respective jurisdictions as follows: Gale A. Norton of Colorado, Robert A. Butterworth of Florida, Warren Price III of Hawaii, Hubert H. Humphrey III of Minnesota, Donald Stenberg of Nebraska, Robert Del Tufo of New Jersey, Nicholas J. Spaeth of North Dakota, Ernest D. Preate, Jr., of Pennsylvania, Hector Rivera-Cruz of Puerto Rico, James E. O'Neil of Rhode Island, T. Travis Medlock of South Carolina, and Joseph B. Meyer of Wyoming; for the State of Connecticut by Richard Blumenthal, Attorney General, and Arnold B. Feigin and Daniel R. Schaefer, Assistant Attorneys General; for the State of Vermont, Office of Court Administrator, by William B. Gray; for the Missouri Bar by Karen M. Iverson and Timothy K. McNamara; for the National Governors Association et al. by Richard Ruda, Michael J. Wahoske, and Mark B. Rotenberg; and for the Washington Legal Foundation by John C. Cozad, W. Dennis Cross, R. Christopher Abele, Daniel J. Popeo, and John C. Scully.

Daniel G. Spraul filed a brief for Judge John W. Keefe as amicus curiae.

[2] JUSTICE WHITE believes that the "political function" cases are inapposite because they involve limitations on "judicially created scrutiny" rather than "Congress' legislative authority," which is at issue here. Post, at 477. He apparently suggests that Congress has greater authority to interfere with state sovereignty when acting pursuant to its Commerce Clause powers than this Court does when applying the Fourteenth Amendment. Elsewhere in his opinion, JUSTICE WHITE emphasizes that the Fourteenth Amendment was designed as an intrusion on state sovereignty. See post, at 480. That being the case, our diminished scrutiny of state laws in the "political function" cases, brought under the Fourteenth Amendment, argues strongly for special care when interpreting alleged congressional intrusions into state sovereignty under the Commerce Clause.

[3] In EEOC v. Wyoming, 460 U. S. 226 (1983), we held that the extension of the ADEA to the States was a valid exercise of congressional power under the Commerce Clause. We left open, however, the issue whether it was also a valid exercise of Congress' power under § 5 of the Fourteenth Amendment. Cf. Fitzpatrick v. Bitzer, 427 U. S. 445, 453, n. 9 (1976) (extension of Title VII of Civil Rights Act of 1964 to States was pursuant to Congress' § 5 power). Although we need not resolve the issue in this case, I note that at least two Courts of Appeals have held that the ADEA was enacted pursuant to Congress' § 5 power. See Heiar v. Crawford County, 746 F. 2d 1190, 1193-1194 (CA7 1984); Ramirez v. Puerto Rico Fire Service, 715 F. 2d 694, 700 (CA1 1983).

[4] Most of the lower courts that have addressed the issue have concluded that appointed state judges fall within the "appointee[s] on the policymaking level" exception. See 898 F. 2d 598 (CA8 1990) (case below); EEOC v. Massachusetts, 858 F. 2d 52 (CA1 1988); Sabo v. Casey, 757 F. Supp. 587 (ED Pa. 1991); In re Stout, 521 Pa. 571, 559 A. 2d 489 (1989); see also EEOC v. Illinois, 721 F. Supp. 156 (ND Ill. 1989). But see EEOC v. Vermont, 904 F. 2d 794 (CA2 1990); Schlitz v. Virginia, 681 F. Supp. 330 (ED Va.), rev'd on other grounds, 854 F. 2d 43 (CA4 1988).

[5] The dissent argues that we should defer to the EEOC's view regarding the scope of the "policymaking level" exception. See post, at 493-494. I disagree. The EEOC's position is not embodied in any formal issuance from the agency, such as a regulation, guideline, policy statement, or administrative adjudication. Instead, it is merely the EEOC's litigating position in recent lawsuits. Accordingly, it is entitled to little if any deference. See, e. g., Bowen v. Georgetown Univ. Hospital, 488 U. S. 204, 212-213 (1988); St. Agnes Hospital v. Sullivan, 284 U. S. App. D. C. 396, 401, 905 F. 2d 1563, 1568 (1990). Although the dissent does cite to an EEOC decision involving the policymaking exception in Title VII, see post, at 494, that decision did not state, even in dicta, that the exception is limited to those who work closely with elected officials. Rather, it merely stated that the exception applies to officials "on the highest levels of state or local government." CCH EEOC Decisions (1983) ¶ 6725. In any event, the EEOC's position is, for the reasons discussed above, inconsistent with the plain language of the statute at issue. "[N]o deference is due to agency interpretations at odds with the plain language of the statute itself." Public Employees Retirement System of Ohio v. Betts, 492 U. S. 158, 171 (1989).

[6] JUSTICE WHITE finds the dictionary definition of "policymaker" broad enough to include the Missouri judges involved in this case, because judges resolve disputes by choosing "`from among alternatives' and elaborate their choices in order `to guide and . . . determine present and future decisions.'" Ante, at 482. See also 898 F. 2d 598, 601 (CA8 1990) (case below), quoting EEOC v. Massachusetts, 858 F. 2d 52, 55 (CA1 1988). I hesitate to classify judges as policymakers, even at this level of abstraction. Although some part of a judge's task may be to fill in the interstices of legislative enactments, the primary task of a judicial officer is to apply rules reflecting the policy choices made by, or on behalf of, those elected to legislative and executive positions. A judge is first and foremost one who resolves disputes, and not one charged with the duty to fashion broad policies establishing the rights and duties of citizens. That task is reserved primarily for legislators. See EEOC v. Vermont, 904 F. 2d 794, 800-801 (CA2 1990).

Nor am I persuaded that judges should be considered policymakers because they sometimes fashion court rules and are otherwise involved in the administration of the state judiciary. See In re Stout, 521 Pa. 571, 583-586, 559 A. 2d 489, 495-497 (1989). These housekeeping tasks are at most ancillary to a judge's primary function described above.

[7] I disagree with JUSTICE WHITE'S suggestion that this reading of the policymaking exclusion renders it superfluous. Ante, at 483. There exist policymakers who work closely with an appointing official but who are appropriately classified as neither members of his "personal staff" nor "immediate adviser[s] with respect to the exercise of the constitutional or legal powers of the office." Among others, certain members of the Governor's Cabinet and high level state agency officials well might be covered by the policymaking exclusion, as I construe it.

[8] The majority acknowledges this anomaly by noting that "`appointee [on] the policymaking level,' particularly in the context of the other exceptions that surround it, is an odd way for Congress to exclude judges; a plain statement that judges are not `employees' would seem the most efficient phrasing." Ante, at 467. The majority dismisses this objection not by refuting it, but by noting that "we are not looking for a plain statement that judges are excluded." Ibid. For the reasons noted in Part I of JUSTICE WHITE'S opinion, this reasoning is faulty; appointed judges are covered unless they fall within the enumerated exclusions.

[9] Relying on Bowen v. Georgetown Univ. Hospital, 488 U. S. 204 (1988), JUSTICE WHITE would conclude that the EEOC's view of the scope of the policymaking exclusion is entitled to "little if any deference" because it is "merely the EEOC's litigating position in recent lawsuits." Ante, at 485, n. 3. This case is distinguishable from Bowen, however, in two important respects. First, unlike in Bowen, where the Court declined to defer "to agency litigating positions that are wholly unsupported by regulations, rulings, or administrative practice," 488 U. S., at 212, the EEOC here has issued an administrative ruling construing Title VII's cognate policymaking exclusion that is entirely consistent with the agency's subsequent "litigation position" that appointed judges are not the kind of officials on the policymaking level whom Congress intended to exclude from ADEA coverage. See CCH EEOC Decisions (1983) ¶ 6725. Second, the Court in Bowen emphasized that the agency had failed to offer "a reasoned and consistent view of the scope of" the relevant statute and had proffered an interpretation of the statute that was "contrary to the narrow view of that provision advocated in past cases." See 488 U. S., at 212-213. In contrast, however, the EEOC never has wavered from its view that the policymaking exclusion does not apply to appointed judges. Thus, this simply is not a case in which a court is asked to defer to "nothing more than an agency's convenient litigating position." Id., at 213. For all the reasons that deference was inappropriate in Bowen, it is appropriate here.

[10] Because I conclude that the challenged Missouri constitutional provision violates the ADEA, I need not consider petitioners' alternative argument that the mandatory retirement provision violates the Fourteenth Amendment to the United States Constitution. See Carnival Cruise Lines, Inc. v. Shute, 499 U. S. 585, 589-590 (1991).

[491] Solicitor General McCree argued the cause for petitioner. With him on the briefs were Kenneth S. Geller, John S. Irving, Carl L. Taylor, Norton J. Come, and Carol A. De Deo.

Don H. Reuben argued the cause for respondents. With him on the brief were Lawrence Gunnels, James A. Serritella, James A. Klenk, and Jerome J. O'Dowd.^[*]

Briefs of amici curiae urging affirmance were filed by Leo Pfeffer and Earl W. Trent, Jr., for the Baptist Joint Committee on Public Affairs; by Thomas Stephen Neuberger for the Center for Law and Religious Freedom of the Christian Legal Society; by Warren L. Johns, Walter E. Carson, Lee Boothby, and Robert J. Hickey for the General Conference of Seventh-Day Adventists; and by David Goldberger and Barbara P. O'Toole for the Roger Baldwin Foundation of the American Civil Liberties Union, Inc., Illinois Division.

Briefs of amici curiae were filed by Lawrence A. Poltrock and Bruce E. Endy for the American Federation of Teachers (AFL-CIO); by Sharp Whitmore for certain Catholic High Schools in the Archdiocese of Los Angeles and the Diocese of Orange; and by George E. Reed and Patrick F. Geary for the United States Catholic Conference.

This case arises out of the National Labor Relations Board's exercise of jurisdiction over lay faculty members at two groups of Catholic high schools. We granted certiorari to consider two questions: (a) Whether teachers in schools operated by a church to teach both religious and secular subjects are within the jurisdiction granted by the National Labor Relations Act; and (b) if the Act authorizes such jurisdiction, does its exercise violate the guarantees of the Religion Clauses of the First Amendment? 434 U. S. 1061 (1978).

One group of schools is operated by the Catholic Bishop of Chicago, a corporation sole; the other group is operated by the Diocese of Fort Wayne-South Bend, Inc. The group operated by the Catholic Bishop of Chicago consists of two schools, Quigley North and Quigley South.^[1] Those schools are termed "minor seminaries" because of their role in educating high school students who may become priests. At one time, only students who manifested a positive and confirmed desire to be priests were admitted to the Quigley schools. In 1970, the requirement was changed so that students admitted to these schools need not show a definite inclination toward the priesthood. Now the students need only be recommended by their parish priest as having a potential for the priesthood or for Christian leadership. The schools continue to provide special religious instruction not offered in other Catholic secondary schools. The Quigley schools also offer essentially the same college-preparatory curriculum as public secondary schools. Their students participate in a variety of extracurricular activities which include secular as well as religious events. The schools are recognized by the State and accredited by a regional educational organization.^[2]

The Diocese of Fort Wayne-South Bend, Inc., has five high schools.^[3] Unlike the Quigley schools, the special recommendation [493] of a priest is not a prerequisite for admission. Like the Quigley schools, however, these high schools seek to provide a traditional secular education but oriented to the tenets of the Roman Catholic faith; religious training is also mandatory. These schools are similarly certified by the State.^[4]

In 1974 and 1975, separate representation petitions were filed with the Board by interested union organizations for both the Quigley and the Fort Wayne-South Bend schools; representation was sought only for lay teachers.^[5] The schools challenged the assertion of jurisdiction on two grounds: (a) that they do not fall within the Board's discretionary jurisdictional criteria; and (b) that the Religion Clauses of the First Amendment preclude the Board's jurisdiction. The Board rejected the jurisdictional arguments on the basis of its decision in Roman Catholic Archdiocese of Baltimore, 216 N. L. R. B. 249 (1975). There the Board explained that its policy was to decline jurisdiction over religiously sponsored organizations "only when they are completely religious, not just religiously associated." Id., at 250. Because neither group of schools was found to fall within the Board's "completely religious" category, the Board ordered elections. Catholic Bishop of Chicago, 220 N. L. R. B. 359 (1975).^[6]

[494] In the Board-supervised election at the Quigley schools, the Quigley Education Alliance, a union affiliated with the Illinois Education Association, prevailed and was certified as the exclusive bargaining representative for 46 lay teachers. In the Diocese of Fort Wayne-South Bend, the Community Alliance for Teachers of Catholic High Schools, a similar union organization, prevailed and was certified as the representative for the approximately 180 lay teachers. Notwithstanding the Board's order, the schools declined to recognize the unions or to bargain. The unions filed unfair labor practice complaints with the Board under §§ 8 (a) (1) and (5) of the National Labor Relations Act, 49 Stat. 452, as amended, 29 U. S. C. §§ 158 (a) (1) and (5). The schools opposed the General Counsel's motion for summary judgment, again challenging the Board's exercise of jurisdiction over religious schools on both statutory and constitutional grounds.

The Board reviewed the record of previous proceedings and concluded that all of the arguments had been raised or could have been raised in those earlier proceedings. Since the arguments had been rejected previously, the Board granted summary judgment, holding that it had properly exercised its statutory discretion in asserting jurisdiction over these schools.^[7] The Board concluded that the schools had violated the Act and ordered that they cease their unfair labor practices and that they bargain collectively with the unions. Catholic [495] Bishop of Chicago, 224 N. L. R. B. 1221 (1976); Diocese of Fort Wayne-South Bend, Inc., 224 N. L. R. B. 1226 (1976).

The schools challenged the Board's orders in petitions to the Court of Appeals for the Seventh Circuit. That court denied enforcement of the Board's orders. 559 F. 2d 1112 (1977).^[8] The court considered the Board's actions in relation to its discretion in choosing to extend its jurisdiction only to religiously affiliated schools that were not "completely religious." It concluded that the Board had not properly exercised its discretion, because the Board's distinction between "completely religious" and "merely religiously associated" failed to provide a workable guide for the exercise of discretion:

"We find the standard itself to be a simplistic black or white, purported rule containing no borderline demarcation of where `completely religious' takes over or, on the other hand, ceases. In our opinion the dichotomous `completely religious—merely religiously associated' standard provides no workable guide to the exercise of discretion. The determination that an institution is so completely a religious entity as to exclude any viable secular components obviously implicates very sensitive questions of faith and tradition. See, e. g., [Wisconsin v.] Yoder, . . . 406 U. S. 205 [(1972)]." Id., at 1118.

The Court of Appeals recognized that the rejection of the Board's policy as to church-operated schools meant that the Board would extend its jurisdiction to all church-operated [496] schools. The court therefore turned to the question of whether the Board could exercise that jurisdiction, consistent with constitutional limitations. It concluded that both the Free Exercise Clause and the Establishment Clause of the First Amendment foreclosed the Board's jurisdiction. It reasoned that from the initial act of certifying a union as the bargaining agent for lay teachers the Board's action would impinge upon the freedom of church authorities to shape and direct teaching in accord with the requirements of their religion. It analyzed the Board's action in this way:

"At some point, factual inquiry by courts or agencies into such matters [separating secular from religious training] would almost necessarily raise First Amendment problems. If history demonstrates, as it does, that Roman Catholics founded an alternative school system for essentially religious reasons and continued to maintain them as an `integral part of the religious mission of the Catholic Church,' Lemon [v. Kurtzman, 403 U. S. 602], 616 [(1971)], courts and agencies would be hard pressed to take official or judicial notice that these purposes were undermined or eviscerated by the determination to offer such secular subjects as mathematics, physics, chemistry, and English literature." Ibid.

The court distinguished local regulations which required fire inspections or state laws mandating attendance, reasoning that they did not "have the clear inhibiting potential upon the relationship between teachers and employers with which the present Board order is directly concerned." Id., at 1124. The court held that interference with management prerogatives, found acceptable in an ordinary commercial setting, was not acceptable in an area protected by the First Amendment. "The real difficulty is found in the chilling aspect that the requirement of bargaining will impose on the exercise of the bishops' control of the religious mission of the schools." Ibid.

The Board's assertion of jurisdiction over private schools is, as we noted earlier, a relatively recent development. Indeed, in 1951 the Board indicated that it would not exercise jurisdiction over nonprofit, educational institutions because to do so would not effectuate the purposes of the Act. Trustees of Columbia University in the City of New York, 97 N. L. R. B. 424. In 1970, however, the Board pointed to what it saw as an increased involvement in commerce by educational institutions and concluded that this required a different position on jurisdiction. In Cornell University, 183 N. L. R. B. 329, the Board overruled its Columbia University decision. Cornell University was followed by the assertion of jurisdiction over nonprofit, private secondary schools. Shattuck School, 189 N. L. R. B. 886 (1971). See also Judson School, 209 N. L. R. B. 677 (1974). The Board now asserts jurisdiction over all private, nonprofit, educational institutions with gross annual revenues that meet its jurisdictional requirements whether they are secular or religious. 29 CFR § 103.1 (1978). See, e. g., Academia San Jorge, 234 N. L. R. B. 1181 (1978) (advisory opinion stating that Board would not assert jurisdiction over Catholic educational institution which did not meet jurisdictional standards); Windsor School, Inc., 199 N. L. R. B. 457, 200 N. L. R. B. 991 (1972) (declining jurisdiction where private, proprietary school did not meet jurisdictional amounts).

That broad assertion of jurisdiction has not gone unchallenged. But the Board has rejected the contention that the Religion Clauses of the First Amendment bar the extension of its jurisdiction to church-operated schools. Where the Board has declined to exercise jurisdiction, it has done so only on the grounds of the employer's minimal impact on commerce. Thus, in Association of Hebrew Teachers of Metropolitan Detroit, 210 N. L. R. B. 1053 (1974), the Board did not assert jurisdiction over the Association which offered [498] courses in Jewish culture in after-school classes, a nursery school, and a college. The Board termed the Association an "isolated instance of [an] atypical employer." Id., at 1058-1059. It explained: "Whether an employer falls within a given `class' of enterprise depends upon those of its activities which are predominant and give the employing enterprise its character. . . . [T]he fact that an employer's activity. . . is dedicated to a sectarian religious purpose is not a sufficient reason for the Board to refrain from asserting jurisdiction." Id., at 1058. Cf. Board of Jewish Education of Greater Washington, D. C., 210 N. L. R. B. 1037 (1974). In the same year the Board asserted jurisdiction over an Association chartered by the State of New York to operate diocesan high schools. Henry M. Hald High School Assn., 213 N. L. R. B. 415 (1974). It rejected the argument that its assertion of jurisdiction would produce excessive governmental entanglement with religion. In the Board's view, the Association had chosen to entangle itself with the secular world when it decided to hire lay teachers. Id., at 418 n. 7.^[9]

When it ordered an election for the lay professional employees at five parochial high schools in Baltimore in 1975, the Board reiterated its belief that exercise of its jurisdiction is not contrary to the First Amendment:

"[T]he Board's policy in the past has been to decline jurisdiction over similar institutions only when they are completely religious, not just religiously associated, and the Archdiocese concedes that instruction is not limited to religious subjects. That the Archdiocese seeks to provide an education based on Christian principles does not lead to a contrary conclusion. Most religiously associated institutions seek to operate in conformity with [499] their religious tenets." Roman Catholic Archdiocese of Baltimore, 216 N. L. R. B., at 250.

The Board also rejected the First Amendment claims in Cardinal Timothy Manning, Roman Catholic Archbishop of the Archdiocese of Los Angeles, 223 N. L. R. B. 1218, 1218 (1976): "Regulation of labor relations does not violate the First Amendment when it involves a minimal intrusion on religious conduct and is necessary to obtain [the Act's] objective." (Emphasis added.)

The Board thus recognizes that its assertion of jurisdiction over teachers in religious schools constitutes some degree of intrusion into the administration of the affairs of church-operated schools. Implicit in the Board's distinction between schools that are "completely religious" and those "religiously associated" is also an acknowledgment of some degree of entanglement. Because that distinction was measured by a school's involvement with commerce, however, and not by its religious association, it is clear that the Board never envisioned any sort of religious litmus test for determining when to assert jurisdiction. Nevertheless, by expressing its traditional jurisdictional standards in First Amendment terms, the Board has plainly recognized that intrusion into this area could run afoul of the Religion Clauses and hence preclude jurisdiction on constitutional grounds.

That there are constitutional limitations on the Board's actions has been repeatedly recognized by this Court even while acknowledging the broad scope of the grant of jurisdiction. The First Amendment, of course, is a limitation on the power of Congress. Thus, if we were to conclude that the Act granted the challenged jurisdiction over these teachers we would be required to decide whether that was constitutionally permissible under the Religion Clauses of the First Amendment.

[500] Although the respondents press their claims under the Religion Clauses, the question we consider first is whether Congress intended the Board to have jurisdiction over teachers in church-operated schools. In a number of cases the Court has heeded the essence of Mr. Chief Justice Marshall's admonition in Murray v. The Charming Betsy, 2 Cranch 64, 118 (1804), by holding that an Act of Congress ought not be construed to violate the Constitution if any other possible construction remains available. Moreover, the Court has followed this policy in the interpretation of the Act now before us and related statutes.

In Machinists v. Street, 367 U. S. 740 (1961), for example, the Court considered claims that serious First Amendment questions would arise if the Railway Labor Act were construed to allow compulsory union dues to be used to support political candidates or causes not approved by some members. The Court looked to the language of the Act and the legislative history and concluded that they did not permit union dues to be used for such political purposes, thus avoiding "serious doubt of [the Act's] constitutionality." Id., at 749.

Similarly in McCulloch v. Sociedad Nacional de Marineros de Honduras, 372 U. S. 10 (1963), a case involving the Board's assertion of jurisdiction over foreign seamen, the Court declined to read the National Labor Relations Act so as to give rise to a serious question of separation of powers which in turn would have implicated sensitive issues of the authority of the Executive over relations with foreign nations. The international implications of the case led the Court to describe it as involving "public questions particularly high in the scale of our national interest." Id., at 17. Because of those questions the Court held that before sanctioning the Board's exercise of jurisdiction " `there must be present the affirmative intention of the Congress clearly expressed.' " Id., at 21-22 (quoting Benz v. Compania Naviera Hidalgo, 353 U. S. 138, 147 (1957)).

[501] The values enshrined in the First Amendment plainly rank high "in the scale of our national values." In keeping with the Court's prudential policy it is incumbent on us to determine whether the Board's exercise of its jurisdiction here would give rise to serious constitutional questions. If so, we must first identify "the affirmative intention of the Congress clearly expressed" before concluding that the Act grants jurisdiction.

In recent decisions involving aid to parochial schools we have recognized the critical and unique role of the teacher in fulfilling the mission of a church-operated school. What was said of the schools in Lemon v. Kurtzman, 403 U. S. 602, 617 (1971), is true of the schools in this case: "Religious authority necessarily pervades the school system." The key role played by teachers in such a school system has been the predicate for our conclusions that governmental aid channeled through teachers creates an impermissible risk of excessive governmental entanglement in the affairs of the church-operated schools. For example, in Lemon, supra, at 617, we wrote:

"In terms of potential for involving some aspect of faith or morals in secular subjects, a textbook's content is ascertainable, but a teacher's handling of a subject is not. We cannot ignore the danger that a teacher under religious control and discipline poses to the separation of the religious from the purely secular aspects of pre-college education. The conflict of functions inheres in the situation." (Emphasis added.)

Only recently we again noted the importance of the teacher's function in a church school: "Whether the subject is `remedial reading,' `advanced reading,' or simply `reading,' a teacher remains a teacher, and the danger that religious doctrine will become intertwined with secular instruction persists." Meek v. Pittenger, 421 U. S. 349, 370 (1975). Cf. [502] Wolman v. Walter, 433 U. S. 229, 244 (1977). Good intentions by government—or third parties—can surely no more avoid entanglement with the religious mission of the school in the setting of mandatory collective bargaining than in the well-motivated legislative efforts consented to by the church-operated schools which we found unacceptable in Lemon, Meek, and Wolman.

The Board argues that it can avoid excessive entanglement since it will resolve only factual issues such as whether an anti-union animus motivated an employer's action. But at this stage of our consideration we are not compelled to determine whether the entanglement is excessive as we would were we considering the constitutional issue. Rather, we make a narrow inquiry whether the exercise of the Board's jurisdiction presents a significant risk that the First Amendment will be infringed.

Moreover, it is already clear that the Board's actions will go beyond resolving factual issues. The Court of Appeals' opinion refers to charges of unfair labor practices filed against religious schools. 559 F. 2d, at 1125, 1126. The court observed that in those cases the schools had responded that their challenged actions were mandated by their religious creeds. The resolution of such charges by the Board, in many instances, will necessarily involve inquiry into the good faith of the position asserted by the clergy-administrators and its relationship to the school's religious mission. It is not only the conclusions that may be reached by the Board which may impinge on rights guaranteed by the Religion Clauses, but also the very process of inquiry leading to findings and conclusions.^[10]

The Board's exercise of jurisdiction will have at least one other impact on church-operated schools. The Board will be called upon to decide what are "terms and conditions of [503] employment" and therefore mandatory subjects of bargaining. See 29 U. S. C. § 158 (d). Although the Board has not interpreted that phrase as it relates to educational institutions, similar state provisions provide insight into the effect of mandatory bargaining. The Oregon Court of Appeals noted that "nearly everything that goes on in the schools affects teachers and is therefore arguably a `condition of employment.' " Springfield Education Assn. v. Springfield School Dist. No. 19, 24 Ore. App. 751, 759, 547 P. 2d 647, 650 (1976).

The Pennsylvania Supreme Court aptly summarized the effect of mandatory bargaining when it observed that the "introduction of a concept of mandatory collective bargaining, regardless of how narrowly the scope of negotiation is defined, necessarily represents an encroachment upon the former autonomous position of management." Pennsylvania Labor Relations Board v. State College Area School Dist., 461 Pa. 494, 504, 337 A. 2d 262, 267 (1975). Cf. Clark County School Dist. v. Local Government Employee-Management Relations Board, 90 Nev. 442, 447, 530 P. 2d 114, 117-118 (1974). See M. Lieberman & M. Moskow, Collective Negotiations for Teachers 221-247 (1966). Inevitably the Board's inquiry will implicate sensitive issues that open the door to conflicts between clergy-administrators and the Board, or conflicts with negotiators for unions. What we said in Lemon, supra, at 616, applies as well here:

"[P]arochial schools involve substantial religious activity and purpose.

"The substantial religious character of these church-related schools gives rise to entangling church-state relationships of the kind the Religion Clauses sought to avoid." (Footnote omitted.)

Mr. Justice Douglas emphasized this in his concurring opinion in Lemon, noting "the admitted and obvious fact that the raison d'êetre of parochial schools is the propagation of a religious faith." 403 U.S., at 628.

[504] The church-teacher relationship in a church-operated school differs from the employment relationship in a public or other nonreligious school. We see no escape from conflicts flowing from the Board's exercise of jurisdiction over teachers in church-operated schools and the consequent serious First Amendment questions that would follow. We therefore turn to an examination of the National Labor Relations Act to decide whether it must be read to confer jurisdiction that would in turn require a decision on the constitutional claims raised by respondents.

There is no clear expression of an affirmative intention of Congress that teachers in church-operated schools should be covered by the Act. Admittedly, Congress defined the Board's jurisdiction in very broad terms; we must therefore examine the legislative history of the Act to determine whether Congress contemplated that the grant of jurisdiction would include teachers in such schools.

In enacting the National Labor Relations Act in 1935, Congress sought to protect the right of American workers to bargain collectively. The concern that was repeated throughout the debates was the need to assure workers the right to organize to counterbalance the collective activities of employers which had been authorized by the National Industrial Recovery Act. But congressional attention focused on employment in private industry and on industrial recovery. See, e. g., 79 Cong. Rec. 7573 (1935) (remarks of Sen. Wagner), 2 National Labor Relations Board, Legislative History of the National Labor Relations Act, 1935, pp. 2341-2343 (1949).

Our examination of the statute and its legislative history indicates that Congress simply gave no consideration to church-operated schools. It is not without significance, however, that the Senate Committee on Education and Labor chose a college professor's dispute with the college as an example of [505] employer-employee relations not covered by the Act. S. Rep. No. 573, 74th Cong., 1st Sess., 7 (1935), 2 Legislative History, supra, at 2307.

Congress' next major consideration of the jurisdiction of the Board came during the passage of the Labor Management Relations Act of 1947—the Taft-Hartley Act. In that Act Congress amended the definition of "employer" in § 2 of the original Act to exclude nonprofit hospitals. 61 Stat. 137, 29 U. S. C. § 152 (2) (1970 ed.). There was some discussion of the scope of the Board's jurisdiction but the consensus was that nonprofit institutions in general did not fall within the Board's jurisdiction because they did not affect commerce. See H. R. 3020, 80th Cong., 1st Sess. (1947), 1 National Labor Relations Board, Legislative History of the Labor Management Relations Act, 1947, p. 34 (1948) (hereinafter Leg. Hist.); H. R. Rep. No. 245, 80th Cong., 1st Sess., 12 (1947), 1 Leg. Hist. 303; H. R. Conf. Rep. No. 510, 80th Cong., 1st Sess., 3, 32 (1947), 1 Leg. Hist. 507, 536; 93 Cong. Rec. 4997 (1947), 2 Leg. Hist. 1464 (remarks of Sens. Tydings and Taft).^[11]

The most recent significant amendment to the Act was passed in 1974, removing the exemption of nonprofit hospitals. Pub. L. 93-360, 88 Stat. 395. The Board relies upon that amendment as showing that Congress approved the Board's exercise of jurisdiction over church-operated schools. A close examination of that legislative history, however, reveals nothing to indicate an affirmative intention that such schools be within the Board's jurisdiction. Since the Board did not assert jurisdiction over teachers in a church-operated [506] school until after the 1974 amendment, nothing in the history of the amendment can be read as reflecting Congress' tacit approval of the Board's action.

During the debate there were expressions of concern about the effect of the bill on employees of religious hospitals whose religious beliefs would not permit them to join a union. 120 Cong. Rec. 12946, 16914 (1974), Legislative History of the Coverage of Nonprofit Hospitals under the National Labor Relations Act, 1974, 93d Cong., 2d Sess., 118, 331-332 (1974) (remarks of Sen. Ervin and Rep. Erlenborn). The result of those concerns was an amendment which reflects congressional sensitivity to First Amendment guarantees:

"Any employee of a health care institution who is a member of and adheres to established and traditional tenets or teachings of a bona fide religion, body, or sect which has historically held conscientious objections to joining or financially supporting labor organizations shall not be required to join or financially support any labor organization as a condition of employment; except that such employee may be required, in lieu of periodic dues and initiation fees, to pay sums equal to such dues and initiation fees to a nonreligious charitable fund exempt from taxation under section 501 (c) (3) of title 26, chosen by such employee from a list of at least three such funds, designated in a contract between such institution and a labor organization, or if the contract fails to designate such funds, then to any such fund chosen by the employee." 29 U. S. C. § 169.

The absence of an "affirmative intention of the Congress clearly expressed" fortifies our conclusion that Congress did not contemplate that the Board would require church-operated schools to grant recognition to unions as bargaining agents for their teachers.

The Board relies heavily upon Associated Press v. NLRB, [507] 301 U. S. 103 (1937). There the Court held that the First Amendment was no bar to the application of the Act to the Associated Press, an organization engaged in collecting information and news throughout the world and distributing it to its members. Perceiving nothing to suggest that application of the Act would infringe First Amendment guarantees of press freedoms, the Court sustained Board jurisdiction. Id., at 131-132. Here, on the contrary, the record affords abundant evidence that the Board's exercise of jurisdiction over teachers in church-operated schools would implicate the guarantees of the Religion Clauses.

Accordingly, in the absence of a clear expression of Congress' intent to bring teachers in church-operated schools within the jurisdiction of the Board, we decline to construe the Act in a manner that could in turn call upon the Court to resolve difficult and sensitive questions arising out of the guarantees of the First Amendment Religion Clauses.

Affirmed.

Q. [by Hearing Officer] Now, we have had quite a bit of testimony already as to liturgies, and I don't want to beat a dead horse; but let me ask you one question: If you know, how many liturgies are required at Catholic parochial high schools; do you know?

A. I think our first problem with that would be defining liturgies. That word would have many definitions. Do you want to go into that?

Q. I believe you defined it before, is that correct, when you first testified?

A. I am not sure. Let me try briefly to do it again, okay?

Q. Yes.

A. A liturgy can range anywhere from the strictest sense of the word, which is the sacrifice of the Mass in the Roman [508] Catholic terminology. It can go from that all the way down to a very informal group in what we call shared prayer.

Two or three individuals praying together and reflecting their own reactions to a scriptural reading. All of these—and there is a big spectrum in between those two extremes—all of these are popularly referred to as liturgies.

Q. I see.

A. Now, possibly in repeating your question, you could give me an idea of that spectrum, I could respond more accurately.

Q. Well, let us stick with the formal Masses. If you know, how many Masses are required at Catholic parochial high schools?

A. Some have none, none required. Some would have two or three during the year where what we call Holy Days of Obligation coincide with school days. Some schools on those days prefer to have a Mass within the school day so the students attend there, rather than their parish churches. Some schools feel that is not a good idea; they should always be in their parish church; so that varies a great deal from school to school.

The Court today holds that coverage of the National Labor Relations Act does not extend to lay teachers employed by church-operated schools. That construction is plainly wrong in light of the Act's language, its legislative history, and this Court's precedents. It is justified solely on the basis of a canon of statutory construction seemingly invented by the Court for the purpose of deciding this case. I dissent.

The general principle of construing statutes to avoid unnecessary constitutional decisions is a well-settled and salutary [509] one. The governing canon, however, is not that expressed by the Court today. The Court requires that there be a "clear expression of an affirmative intention of Congress" before it will bring within the coverage of a broadly worded regulatory statute certain persons whose coverage might raise constitutional questions. Ante, at 504. But those familiar with the legislative process know that explicit expressions of congressional intent in such broadly inclusive statutes are not commonplace. Thus, by strictly or loosely applying its requirement, the Court can virtually remake congressional enactments. This flouts Mr. Chief Justice Taft's admonition "that amendment may not be substituted for construction, and that a court may not exercise legislative functions to save [a] law from conflict with constitutional limitation." Yu Cong Eng v. Trinidad, 271 U. S. 500, 518 (1926). See Aptheker v. Secretary of State, 378 U. S. 500, 515 (1964); Jay v. Boyd, 351 U. S. 345, 357 n. 21 (1956); Shapiro v. United States, 335 U. S. 1, 31, and n. 40 (1948); United States v. Sullivan, 332 U. S. 689, 693 (1948); Hopkins Savings Assn. v. Cleary, 296 U. S. 315, 335 (1935).^[12]

[510] The settled canon for construing statutes wherein constitutional questions may lurk was stated in Machinists v. Street, 367 U. S. 740 (1961), cited by the Court, ante, at 500:

" `When the validity of an act of the Congress is drawn in question, and even if a serious doubt of constitutionality is raised, it is a cardinal principle that this Court will first ascertain whether a construction of the statute is fairly possible by which the question may be avoided.' Crowell v. Benson, 285 U. S. 22, 62." Id., at 749-750 (emphasis added).^[13]

Accord, Pernell v. Southall Realty, 416 U. S. 363, 365 (1974); Johnson v. Robison, 415 U. S. 361, 367 (1974); Curtis v. Loether, 415 U. S. 189, 192 n. 6 (1974); Ashwander v. TVA, 297 U. S. 288, 348 (1936) (Brandeis, J., concurring); Moore Ice Cream Co. v. Rose, 289 U. S. 373, 379 (1933). This limitation to constructions that are "fairly possible," and "reasonable," see Yu Cong Eng v. Trinidad, supra, at 518, acts as a [511] brake against wholesale judicial dismemberment of congressional enactments. It confines the judiciary to its proper role in construing statutes, which is to interpret them so as to give effect to congressional intention. The Court's new "affirmative expression" rule releases that brake.

The interpretation of the National Labor Relations Act announced by the Court today is not "fairly possible." The Act's wording, its legislative history, and the Court's own precedents leave "the intention of the Congress . . . revealed too distinctly to permit us to ignore it because of mere misgivings as to power." Moore Ice Cream Co. v. Rose, supra, at 379. Section 2 (2) of the Act, 29 U. S. C. § 152 (2), defines "employer" as

". . . any person acting as an agent of an employer, directly or indirectly, but shall not include the United States or any wholly owned Government corporation, or any Federal Reserve Bank, or any State or political subdivision thereof, or any person subject to the Railway Labor Act, as amended from time to time, or any labor organization (other than when acting as an employer), or anyone acting in the capacity of officer or agent of such labor organization." (Emphasis added.)

Thus, the Act covers all employers not within the eight express exceptions. The Court today substitutes amendment for construction to insert one more exception—for church-operated schools. This is a particularly transparent violation of the judicial role: The legislative history reveals that Congress itself considered and rejected a very similar amendment.

The pertinent legislative history of the NLRA begins with the Wagner Act of 1935, 49 Stat. 449. Section 2 (2) of that Act, identical in all relevant respects to the current section, excluded from its coverage neither church-operated schools [512] nor any other private nonprofit organization.^[14] Accordingly, in applying that Act, the National Labor Relations Board did not recognize an exception for nonprofit employers, even when religiously associated.^[15] An argument for an implied nonprofit exemption was rejected because the design of the Act was as clear then as it is now: "[N]either charitable institutions nor their employees are exempted from operation of the Act by its terms, although certain other employers and employees are exempted." Central Dispensary & Emergency Hospital, 44 N. L. R. B. 533, 540 (1942) (footnotes omitted), enf'd, 79 U. S. App. D. C. 274, 145 F. 2d 852 (1944). Both the lower courts and this Court concurred in the Board's construction. See Polish National Alliance v. NLRB, 322 U. S. 643 (1944), aff'g 136 F. 2d 175 (CA7 1943); Associated Press v. NLRB, 301 U. S. 103 (1937), aff'g 85 F. 2d 56 (CA2 1936); NLRB v. Central Dispensary & Emergency Hospital, 79 U. S. App. D. C. 274, 145 F. 2d 852 (1944).

The Hartley bill, which passed the House of Representatives [513] in 1947, would have provided the exception the Court today writes into the statute:

"The term `employer' . . . shall not include . . . any corporation, community chest, fund, or foundation organized and operated exclusively for religious, charitable, scientific, literary, or educational purposes, . . . no part of the net earnings of which inures to the benefit of any private shareholder or individual. . . ." (Emphasis added.) H. R. 3020, 80th Cong., 1st Sess., § 2 (2) (Apr. 18, 1947), reprinted in National Labor Relations Board, Legislative History of the Labor Management Relations Act, 1947, pp. 160-161 (hereinafter, 1947 Leg. Hist.).

But the proposed exception was not enacted.^[16] The bill reported by the Senate Committee on Labor and Public Welfare did not contain the Hartley exception. See S. 1126, 80th Cong., 1st Sess., § 2 (2) (Apr. 17, 1947), 1947 Leg. Hist. 99, 102. Instead, the Senate proposed an exception limited to nonprofit hospitals, and passed the bill in that form. See H. R. 3020, 80th Cong., 1st Sess., § 2 (2) (Senate, May 13, 1947), 1947 Leg. Hist. 226, 229. The Senate version was accepted by the House in conference, thus limiting the exception [514] for nonprofit employers to nonprofit hospitals. Ch. 120, 61 Stat. 136.^[17]

Even that limited exemption was ultimately repealed in 1974. Pub. L. 93-360, 88 Stat. 395. In doing so, Congress confirmed the view of the Act expressed here: that it was intended to cover all employers—including nonprofit employers— unless expressly excluded, and that the 1947 amendment excluded only nonprofit hospitals. See H. R. Rep. No. 93-1051, [515] p. 4 (1974), reprinted in Senate Committee on Labor and Public Welfare, Legislative History of the Coverage of Nonprofit Hospitals under the National Labor Relations Act, 1974, p. 272 (Comm. Print 1974) (hereafter 1974 Leg. Hist.); 120 Cong. Rec. 12938 (1974), 1974 Leg. Hist. 95 (Sen. Williams); 120 Cong. Rec. 16900 (1974), 1974 Leg. Hist. 291 (Rep. Ashbrook).^[18] Moreover, it is significant that in considering the 1974 amendments, the Senate expressly rejected an amendment proposed by Senator Ervin that was analogous to the one the Court today creates—an amendment to exempt nonprofit hospitals operated by religious groups. 120 Cong. Rec. 12950, 12968 (1974), 1974 Leg. Hist. 119, 141. Senator Cranston, floor manager of the Senate Committee bill and primary opponent of the proposed religious exception, explained:

"[S]uch an exception for religiously affiliated hospitals would seriously erode the existing national policy which holds religiously affiliated institutions generally such as proprietary nursing homes, residential communities, and educational facilities to the same standards as their nonsectarian counterparts." 120 Cong. Rec. 12957 (1974), 1974 Leg. Hist. 137 (emphasis added).

[516] See also ibid. (Sen. Javits); 120 Cong. Rec. 12957 (1974), 1974 Leg. Hist. 138 (Sen. Williams).^[19]

In construing the Board's jurisdiction to exclude church-operated schools, therefore, the Court today is faithful to neither the statute's language nor its history. Moreover, it is also untrue to its own precedents. "This Court has consistently declared that in passing the National Labor Relations Act, Congress intended to and did vest in the Board the fullest jurisdictional breadth constitutionally permissible under the Commerce Clause. See, e. g., Guss v. Utah Labor Board, 353 U. S. 1, 3; Polish Alliance v. Labor Board, 322 U. S. 643, 647-648; Labor Board v. Fainblatt, 306 U. S. 601, 607." NLRB v. Reliance Fuel Oil Corp., 371 U. S. 224, 226 (1963) (emphasis in original). As long as an employer is within the reach of Congress' power under the Commerce Clause—and no one doubts that respondents are—the Court has held him to be covered by the Act regardless of the nature of his activity. See, e. g., Polish National Alliance v. NLRB, 322 U. S. 643 (1944) (nonprofit fraternal organization). Indeed, Associated Press v. NLRB, 301 U. S. 103 (1937), construed the Act to [517] cover editorial employees of a nonprofit news-gathering organization despite a claim—precisely parallel to that made here—that their inclusion rendered the Act in violation of the First Amendment.^[20] Today's opinion is simply unable to explain the grounds that distinguish that case from this one.^[21]

Thus, the available authority indicates that Congress intended to include—not exclude—lay teachers of church-operated schools. The Court does not counter this with evidence that Congress did intend an exception it never stated. Instead, despite the legislative history to the contrary, it construes the Act as excluding lay teachers only because Congress did not state explicitly that they were covered. In Mr. Justice Cardozo's words, this presses "avoidance of a [518] difficulty . . . to the point of disingenuous evasion." Moore Ice Cream Co. v. Rose, 289 U. S., at 379.^[22]

Under my view that the NLRA includes within its coverage lay teachers employed by church-operated schools, the constitutional questions presented would have to be reached. I do not now do so only because the Court does not. See Sierra Club v. Morton, 405 U. S. 727, 755 (1972) (BRENNAN, J., dissenting). I repeat for emphasis, however, that while the resolution of the constitutional question is not without difficulty, it is irresponsible to avoid it by a cavalier exercise in statutory interpretation which succeeds only in defying congressional intent. A statute is not "a nose of wax to be changed from that which the plain language imports . . . ." Yu Cong Eng v. Trinidad, 271 U. S., at 518.

----------

[*] J. Albert Woll and Laurence Gold filed a brief for the American Federation of Labor and Congress of Industrial Organizations as amicus curiae urging reversal.

[1] The Catholic Bishop operates other schools in the Chicago area, but they were not involved in the proceedings before the Board.

[2] As explained to the Board's Hearing Officer, in Illinois the term "approval" is distinct from "recognition." Before a school may operate, it must be approved by the State's Department of Education. Approval is given when a school meets the minimal requirements under state law, such as for compulsory attendance; approval does not require any evaluation of the school's program. Recognition, which is not required to operate, is given only after the school has passed the State's evaluation.

[3] The Diocese also has 47 elementary schools. They were not involved in the proceedings before the Board.

[4] As explained to the Board's Hearing Officer, "certification" by the State of Indiana is roughly equivalent to "recognition" by the State of Illinois. Both are voluntary procedures which involve some evaluation by the state educational authorities.

[5] The certification and order cover only "all full-time and regular part-time lay teachers, including physical education teachers . . .; and excluding rectors, procurators, dean of studies, business manager, director of student activities, director of formation, director of counseling services, office clerical employees, maintenance employees, cafeteria workers, watchmen, librarians, nurses, all religious faculty, and all guards and supervisors as defined in the Act . . . ." Catholic Bishop of Chicago, 220 N. L. R. B. 359, 360 (1975).

[6] The decision concerning the Diocese of Fort Wayne-South Bend, Inc., is not reported.

[7] The Board relied on its reasoning in Cardinal Timothy Manning, Roman Catholic Archbishop of the Archdiocese of Los Angeles, 223 N. L. R. B. 1218 (1976): "We also do not agree that the schools are religious institutions intimately involved with the Catholic Church. It has heretofore been the Board's policy to decline jurisdiction over institutions only when they are completely religious, not just religiously associated. Roman Catholic Archdiocese of Baltimore, Archdiocesan High Schools, 216 NLRB 249 (1975). The schools perform in part the secular function of educating children, and in part concern themselves with religious instruction. Therefore, we will not decline to assert jurisdiction over these schools on such a basis." 223 N. L. R. B., at 1218.

[8] Cf. Caulfield v. Hirsch, 95 LRRM 3164 (ED Pa. 1977) (enjoining Board from asserting jurisdiction over elementary schools in Archdiocese of Philadelphia). This case is presently under review by the Court of Appeals for the Third Circuit. See App. to Pet. for Cert. in Caulfield v. Hirsch, O. T. 1977, No. 77-1411, p. A76, cert. denied, 436 U. S. 957 (1978).

[9] The Board went on to explain that the rights guaranteed by § 7 of the Act, 29 U. S. C. § 157, were "a part of our national heritage established by Congress, [and] were a legitimate exercise of Congress' constitutional power." 213 N. L. R. B., at 418 n. 7.

[10] This kind of inquiry and its sensitivity are illustrated in the examination of Monsignor O'Donnell, the Rector of Quigley North, by the Board's Hearing Officer, which is reproduced in the appendix to this opinion.

[11] The National Labor Relations Act was amended again when Congress passed the Labor-Management Reporting and Disclosure Act in 1959. 73 Stat. 519. That Act made no changes in the definition of "employer" and the legislative history contains no reference to church-operated schools. See generally National Labor Relations Board, Legislative History of the Labor-Management Reporting and Disclosure Act of 1959 (1959).

[12] The Court's new canon derives from the statement, " `there must be present the affirmative intention of the Congress clearly expressed,' " in McCulloch v. Sociedad Nacional de Marineros de Honduras, 372 U. S. 10, 21-22 (1963). Reliance upon that case here is clearly misplaced. The question in McCulloch was whether the National Labor Relations Act. extended to foreign seamen working aboard foreign-flag vessels. No question as to the constitutional power of Congress to cover foreign crews was presented. Indeed, all parties agreed that Congress was constitutionally empowered to reach the foreign seamen involved while they were in American waters. Id.,at 17. The only question was whether Congress had intended to do so.

The McCulloch Court held that Congress had not meant to reach disputes between foreign shipowners and their foreign crews. McCulloch, however, did not turn simply upon an absence of affirmative evidence that Congress wanted to reach alien seamen, but rather upon the fact, as a prior case had already held, that the legislative history " `inescapably describe[d] the boundaries of the Act as including only the workingmen of our own country and its possessions,' " Id., at 18, quoting Benz v. Compania Naviera Hidalgo, 353 U. S. 138, 144 (1957). The Court also noted that under well-established rules of international law, "the law of the flag state ordinarily governs the internal affairs of a ship. See Wildenhus's Case, [120 U. S. 1,] 12." 372 U. S., at 21. In light of that contrary legislative history and domestic and international precedent, it is not at all surprising that McCulloch balked at holding foreign seamen covered without a strong affirmative showing of congressional intent. As the Court today admits, there is no such contrary legislative history or precedent with respect to jurisdiction over church-operated schools. Ante, at 504. The McCulloch statement, therefore, has no role to play in this case.

[13] In Street, the Court construed the Railway Labor Act as not permitting the use of an employee's compulsorily checked-off union dues for political causes with which he disagreed. As in McCulloch, see n. 1, supra, it so held not because of an absence of affirmative evidence that Congress did mean to permit such uses, but rather because the language and history of the Act indicated affirmatively that Congress did not mean to permit such constitutionally questionable practices. See 367 U. S., at 765-770.

[14] Section 2 (2), 49 Stat. 450, stated:

"The term `employer' includes any person acting in the interest of an employer, directly or indirectly, but shall not include the United States, or any State or political subdivision thereof, or any person subject to the Railway Labor Act, as amended from time to time, or any labor organization (other than when acting as an employer), or anyone acting in the capacity of officer or agent of such labor organization."

[15] See Christian Board of Publication, 13 N. L. R. B. 534, 537 (1939), enf'd, 113 F. 2d 678 (CA8 1940); American Medical Assn., 39 N. L. R. B. 385, 386 (1942); Central Dispensary & Emergency Hospital, 44 N. L. R. B. 533, 539 (1942), enf'd, 79 U. S. App. D. C. 274, 145 F. 2d 852 (1944); Henry Ford Trade School, 58 N. L. R. B. 1535, 1536 (1944); Polish National Alliance, 42 N. L. R. B. 1375, 1380 (1942), enf'd, 136 F. 2d 175 (CA7 1943), aff'd, 322 U. S. 643 (1944); Associated Press, 1 N. L. R. B. 788, 790, enf'd, 85 F. 2d 56 (CA2 1936), aff'd, 301 U. S. 103 (1937). In unpublished decisions, the Board also exercised jurisdiction over the YWCA and the Welfare & Recreational Association. See Central Dispensary & Emergency Hospital, 44 N. L. R. B., at 538 n. 8.

[16] A number of reasons were offered for the rejection of the Hartley bill's exception. Some Congressmen strongly opposed the exception, see 93 Cong. Rec. 3446 (1947) (remarks of Rep. Klein); some were opposed to additional exceptions to the Board's jurisdiction, see id., at 4997 (remarks of Sen. Taft); and some thought it unnecessary, see H. R. Conf. Rep. No. 510, 80th Cong., 1st Sess., 32 (1947), 1947 Leg. Hist. 536. See generally NLRB v. Wentworth Institute, 515 F. 2d 550, 555 (CA1 1975) ("[P]erhaps the most obvious, interpretation of the rejection of the House exclusion would be that Congress meant to include nonprofit organizations [within the scope of the Act]"); Sherman & Black, The Labor Board and the Private Nonprofit Employer: A Critical Examination of the Board's Worthy Cause Exemption, 83 Harv. L. Rev. 1323, 1331-1337 (1970). But whatever the reasons, it is clear that an amendment similar to that made by the Court today was proposed and rejected in 1947.

[17] The Board's contemporaneous construction of the 1947 amendment was that only nonprofit hospitals were intended to be exempt. In 1950, for example, in asserting jurisdiction over a nonprofit religious organization, the Board stated:

"The Employer asserts that, as it is a nonprofit organization which is engaged in purely religious activities, it is not engaged in commerce within the meaning of the Act. We find no merit in this contention. . . . As this Board and the courts have held, it is immaterial that the Employer may be a nonprofit organization, or that its activities may be motivated by considerations other than those applicable to enterprises which are, in the generally accepted sense, commercial." Sunday School Board of the Southern Baptist Convention, 92 N. L. R. B. 801, 802.

It is true that in Trustees of Columbia University, 97 N. L. R. B. 424 (1951), the Board indicated that it would not exercise jurisdiction over nonprofit, educational institutions; but it expressly did so as a matter of discretion, affirming that the activities of the University did come within the Act and the Board's jurisdiction. Id., at 425. That 1951 discretionary decision does not undermine the validity of the Board's determination in Cornell University, 183 N. L. R. B. 329 (1970), that changing conditions—particularly the increasing impact of such institutions on interstate commerce—now required a change in policy leading to the renewed exercise of Board jurisdiction. As we emphasized in NLRB v. Weingarten, Inc., 420 U. S. 251, 265-266 (1975):

"To hold that the Board's earlier decisions froze the development of this important aspect of the national labor law would misconceive the nature of administrative decisionmaking. `"Cumulative experience" begets understanding and insight by which judgments . . . are validated or qualified or invalidated. The constant process of trial and error, on a wider and fuller scale than a single adversary litigation permits, differentiates perhaps more than anything else the administrative from the judicial process.' NLRB v. Seven-Up Co., 344 U.S. 344, 349 (1953)."

[18] The House Report stated: "Currently, the only broad area of charitable, eleemosynary, educational institutions wherein the Board does not now exercise jurisdiction concerns the nonprofit hospitals, explicitly excluded by section 2 (2) of the Act. . . . [T]he bill removes the existing Taft-Hartley exemption in section 2 (2) of the Act. It restores to the employees of nonprofit hospitals the same rights and protections enjoyed by the employees of proprietary hospitals and most all other employees." H. R. Rep. No. 93-1051, p. 4 (1974), 1974 Leg. Hist. 272. Similarly, Senator Williams, Chairman of the Senate Committee on Labor and Public Welfare, criticized the nonprofit-hospital exemption as "not only inconsistent with the protection enjoyed by proprietary hospitals and other types of health care institutions, but it is also inconsistent with the coverage of other nonprofit activities." 120 Cong. Rec. 12938 (1974), 1974 Leg. Hist. 95. See also 120 Cong. Rec. 16900 (1974), 1974 Leg. Hist. 291 (Rep. Ashbrook).

[19] The Court relies upon the fact that the 1974 amendments provided that "[a]ny employee of a health care institution who is a member of . . . a bona fide religion . . . which has historically held conscientious objections to joining . . . labor organizations shall not be required to join . . . any labor organization as a condition of employment . . . ." 29 U. S. C. § 169 (emphasis added). This is, of course, irrelevant to the instant case, as no employee has alleged that he was required to join a union against his religious principles and not even the respondent employers contend that collective bargaining itself is contrary to their religious beliefs. Recognizing this, the Court has limited its inference from the amendment to the proposition that it reflects "congressional sensitivity to First Amendment guarantees." Ante, at 506. This is quite true, but its usefulness as support for the Court's opinion is completely negated by the rejection of the Ervin amendment, see text, supra, which makes clear the balance struck by Congress. While Congress agreed to exclude conscientiously objecting employees, it expressly refused to sanction an exclusion for all religiously affiliated employers.

[20] Associated Pressstated the employer's argument as follows:

"The conclusion which the petitioner draws is that whatever may be the case with respect to employees in its mechanical departments it must have absolute and unrestricted freedom to employ and to discharge those who, like Watson, edit the news, that there must not be the slightest opportunity for any bias or prejudice personally entertained by an editorial employee to color or to distort what he writes, and that the Associated Press cannot be free to furnish unbiased and impartial news reports unless it is equally free to determine for itself the partiality or bias of editorial employees. So it is said that any regulation protective of union activities, or the right collectively to bargain on the part of such employees, is necessarily an invalid invasion of the freedom of the press." 301 U. S., at 131.

[21] The Court would distinguish Associated Press on the ground that there the Court "[p]erceiv[ed] nothing to suggest that application of the Act would infringe First Amendment guarantees . . . [while h]ere, on the contrary, the record affords abundant evidence that the Board's exercise of jurisdiction . . . would implicate the guarantees of the Religion Clauses." Ante, at 507. But this is mere assertion. The Court does not explain why the press' First Amendment problem in Associated Press was any less substantial than the church-supported schools' First Amendment challenge here. In point of fact, the problems raised are of precisely the same difficulty. The Court therefore cannot square its judicial "reconstruction" of the Act in this case with the refusal to rewrite the same Act in Associated Press.

[22] Not even the Court's redrafting of the statute causes all First Amendment problems to disappear. The Court's opinion implies limitation of its exception to church-operated schools. That limitation is doubtless necessary since this Court has already rejected a more general exception for nonprofit organizations. See Polish National Alliance v. NLRB, 322 U. S. 643 (1944). But such an exemption, available only to church-operated schools, generates a possible Establishment Clause question of its own. Walz v. Tax Comm'n, 397 U. S. 664 (1970), does not put that question to rest, for in upholding the property tax exemption for churches there at issue, we emphasized that New York had "not singled out . . . churches as such; rather, it has granted exemption to all houses of religious worship within a broad class of property owned by nonprofit, quasi-public corporations . . . ." Id., at 673. Like the Court, "at this stage of [my] consideration [I am] not compelled to determine whether the [Establishment Clause problem] is [as significant] as [I] would were [I] considering the constitutional issue." Ante, at 502. It is enough to observe that no matter which way the Court turns in interpreting the Act, it cannot avoid constitutional questions.

529 U.S. 120 (2000)

CERTIORARI TO THE UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT

[121] [122] [123] O'Connor, J., delivered the opinion of the Court, in which Rehnquist, C. J., and Scalia, Kennedy, and Thomas, JJ., joined. Breyer, J., filed a dissenting opinion, in which Stevens, Souter, and Ginsburg, JJ., joined, post, p. 161.

Solicitor General Waxman argued the cause for petitioners. With him on the briefs were Acting Assistant Attorney General Ogden, Deputy Solicitor General Kneedler, Deputy Assistant Attorney General Schultz, Irving L. Gornstein, Eugene Thirolf, Douglas Letter, Gerald C. Kell, Chris- [124] tine N. Kohl, Margaret Jane Porter, Karen E. Schifter, and Patricia J. Kaeding.

Richard M. Cooper argued the cause for respondents. With him on the brief for respondent R. J. Reynolds Tobacco Co. was Steven M. Umin. Andrew S. Krulwich, Bert W. Rein, Thomas W. Kirby, and Michael L. Robinson filed a brief for respondent Brown & Williamson Tobacco Corp. Larry B. Sitton filed a brief for respondents United States Tobacco Co. et al. William C. MacLeod filed a brief for respondents National Association of Convenience Stores et al. Peter T. Grossi, Jr., Arthur N. Levine, Jeff Richman, Richard A. Merrill, and Herbert Dym filed a brief for respondents Philip Morris Inc. et al.^[1]

This case involves one of the most troubling public health problems facing our Nation today: the thousands of premature deaths that occur each year because of tobacco use. In 1996, the Food and Drug Administration (FDA), after having expressly disavowed any such authority since its inception, asserted jurisdiction to regulate tobacco products. See 61 Fed. Reg. 44619-45318. The FDA concluded that nicotine is a "drug" within the meaning of the Food, Drug, and Cosmetic Act (FDCA or Act), 52 Stat. 1040, as amended, 21 U. S. C. § 301 et seq., and that cigarettes and smokeless tobacco are "combination products" that deliver nicotine to the body. 61 Fed. Reg. 44397 (1996). Pursuant to this authority, it promulgated regulations intended to reduce tobacco consumption among children and adolescents. Id., at 44615-44618. The agency believed that, because most tobacco consumers begin their use before reaching the age of 18, curbing tobacco use by minors could substantially reduce the prevalence of addiction in future generations and thus the incidence of tobacco-related death and disease. Id., at 44398-44399.

Regardless of how serious the problem an administrative agency seeks to address, however, it may not exercise its authority "in a manner that is inconsistent with the administrative structure that Congress enacted into law." ETSI Pipeline Project v. Missouri, 484 U. S. 495, 517 (1988). And although agencies are generally entitled to deference in the interpretation of statutes that they administer, a reviewing "court, as well as the agency, must give effect to the unambiguously [126] expressed intent of Congress." Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 842-843 (1984). In this case, we believe that Congress has clearly precluded the FDA from asserting jurisdiction to regulate tobacco products. Such authority is inconsistent with the intent that Congress has expressed in the FDCA's overall regulatory scheme and in the tobacco-specific legislation that it has enacted subsequent to the FDCA. In light of this clear intent, the FDA's assertion of jurisdiction is impermissible.

The FDCA grants the FDA, as the designee of the Secretary of Health and Human Services (HHS), the authority to regulate, among other items, "drugs" and "devices." See 21 U. S. C. §§ 321(g)—(h), 393 (1994 ed. and Supp. III). The Act defines "drug" to include "articles (other than food) intended to affect the structure or any function of the body." 21 U. S. C. § 321(g)(1)(C). It defines "device," in part, as "an instrument, apparatus, implement, machine, contrivance, . . . or other similar or related article, including any component, part, or accessory, which is . . . intended to affect the structure or any function of the body." § 321(h). The Act also grants the FDA the authority to regulate so-called "combination products," which "constitute a combination of a drug, device, or biological product." § 353(g)(1). The FDA has construed this provision as giving it the discretion to regulate combination products as drugs, as devices, or as both. See 61 Fed. Reg. 44400 (1996).

On August 11, 1995, the FDA published a proposed rule concerning the sale of cigarettes and smokeless tobacco to children and adolescents. 60 Fed. Reg. 41314-41787. The rule, which included several restrictions on the sale, distribution, and advertisement of tobacco products, was designed to reduce the availability and attractiveness of tobacco products to young people. Id., at 41314. A public comment period followed, during which the FDA received over 700,000 submissions, [127] more than "at any other time in its history on any other subject." 61 Fed. Reg. 44418 (1996).

On August 28, 1996, the FDA issued a final rule entitled "Regulations Restricting the Sale and Distribution of Cigarettes and Smokeless Tobacco to Protect Children and Adolescents." Id., at 44396. The FDA determined that nicotine is a "drug" and that cigarettes and smokeless tobacco are "drug delivery devices," and therefore it had jurisdiction under the FDCA to regulate tobacco products as customarily marketed—that is, without manufacturer claims of therapeutic benefit. Id., at 44397, 44402. First, the FDA found that tobacco products "`affect the structure or any function of the body' " because nicotine "has significant pharmacological effects." Id., at 44631. Specifically, nicotine "exerts psychoactive, or mood-altering, effects on the brain" that cause and sustain addiction, have both tranquilizing and stimulating effects, and control weight. Id., at 44631-44632. Second, the FDA determined that these effects were "intended" under the FDCA because they "are so widely known and foreseeable that [they] may be deemed to have been intended by the manufacturers," id., at 44687; consumers use tobacco products "predominantly or nearly exclusively" to obtain these effects, id., at 44807; and the statements, research, and actions of manufacturers revealed that they "have `designed' cigarettes to provide pharmacologically active doses of nicotine to consumers," id., at 44849. Finally, the agency concluded that cigarettes and smokeless tobacco are "combination products" because, in addition to containing nicotine, they include device components that deliver a controlled amount of nicotine to the body, id., at 45208-45216.

Having resolved the jurisdictional question, the FDA next explained the policy justifications for its regulations, detailing the deleterious health effects associated with tobacco use. It found that tobacco consumption was "the single leading cause of preventable death in the United States." Id., at 44398. According to the FDA, "[m]ore than 400,000 [128] people die each year from tobacco-related illnesses, such as cancer, respiratory illnesses, and heart disease." Ibid. The agency also determined that the only way to reduce the amount of tobacco-related illness and mortality was to reduce the level of addiction, a goal that could be accomplished only by preventing children and adolescents from starting to use tobacco. Id., at 44398-44399. The FDA found that 82% of adult smokers had their first cigarette before the age of 18, and more than half had already become regular smokers by that age. Id., at 44398. It also found that children were beginning to smoke at a younger age, that the prevalence of youth smoking had recently increased, and that similar problems existed with respect to smokeless tobacco. Id., at 44398-44399. The FDA accordingly concluded that if "the number of children and adolescents who begin tobacco use can be substantially diminished, tobacco-related illness can be correspondingly reduced because data suggest that anyone who does not begin smoking in childhood or adolescence is unlikely ever to begin." Id., at 44399.

Based on these findings, the FDA promulgated regulations concerning tobacco products' promotion, labeling, and accessibility to children and adolescents. See id., at 44615-44618. The access regulations prohibit the sale of cigarettes or smokeless tobacco to persons younger than 18; require retailers to verify through photo identification the age of all purchasers younger than 27; prohibit the sale of cigarettes in quantities smaller than 20; prohibit the distribution of free samples; and prohibit sales through self-service displays and vending machines except in adult-only locations. Id., at 44616-44617. The promotion regulations require that any print advertising appear in a black-and-white, text-only format unless the publication in which it appears is read almost exclusively by adults; prohibit outdoor advertising within 1,000 feet of any public playground or school; prohibit the distribution of any promotional items, such as T-shirts or hats, bearing the manufacturer's brand name; and prohibit a [129] manufacturer from sponsoring any athletic, musical, artistic, or other social or cultural event using its brand name. Id., at 44617-44618. The labeling regulation requires that the statement, "A Nicotine-Delivery Device for Persons 18 or Older," appear on all tobacco product packages. Id., at 44617.

The FDA promulgated these regulations pursuant to its authority to regulate "restricted devices." See 21 U. S. C. § 360j(e). The FDA construed § 353(g)(1) as giving it the discretion to regulate "combination products" using the Act's drug authorities, device authorities, or both, depending on "how the public health goals of the act can be best accomplished." 61 Fed. Reg. 44403 (1996). Given the greater flexibility in the FDCA for the regulation of devices, the FDA determined that "the device authorities provide the most appropriate basis for regulating cigarettes and smokeless tobacco." Id., at 44404. Under 21 U. S. C. § 360j(e), the agency may "require that a device be restricted to sale, distribution, or use . . . upon such other conditions as [the FDA] may prescribe in such regulation, if, because of its potentiality for harmful effect or the collateral measures necessary to its use, [the FDA] determines that there cannot otherwise be reasonable assurance of its safety and effectiveness." The FDA reasoned that its regulations fell within the authority granted by § 360j(e) because they related to the sale or distribution of tobacco products and were necessary for providing a reasonable assurance of safety. 61 Fed. Reg. 44405-44407 (1996).

Respondents, a group of tobacco manufacturers, retailers, and advertisers, filed suit in United States District Court for the Middle District of North Carolina challenging the regulations. See Coyne Beahm, Inc. v. FDA, 966 F. Supp. 1374 (1997). They moved for summary judgment on the grounds that the FDA lacked jurisdiction to regulate tobacco products as customarily marketed, the regulations exceeded the FDA's authority under 21 U. S. C. § 360j(e), and the advertising [130] restrictions violated the First Amendment. Second Brief in Support of Plaintiffs' Motion for Summary Judgment in No. 2:95CV00591 (MDNC), in 3 Rec. in No. 97-1604 (CA4), Tab No. 40; Third Brief in Support of Plaintiffs' Motion for Summary Judgment in No. 2:95CV00591 (MDNC), in 3 Rec. in No. 97-1604 (CA4), Tab No. 42. The District Court granted respondents' motion in part and denied it in part. 966 F. Supp., at 1400. The court held that the FDCA authorizes the FDA to regulate tobacco products as customarily marketed and that the FDA's access and labeling regulations are permissible, but it also found that the agency's advertising and promotion restrictions exceed its authority under § 360j(e). Id., at 1380-1400. The court stayed implementation of the regulations it found valid (except the prohibition on the sale of tobacco products to minors) and certified its order for immediate interlocutory appeal. Id., at 1400-1401.

The Court of Appeals for the Fourth Circuit reversed, holding that Congress has not granted the FDA jurisdiction to regulate tobacco products. See 153 F. 3d 155 (1998). Examining the FDCA as a whole, the court concluded that the FDA's regulation of tobacco products would create a number of internal inconsistencies. Id., at 162-167. Various provisions of the Act require the agency to determine that any regulated product is "safe" before it can be sold or allowed to remain on the market, yet the FDA found in its rulemaking proceeding that tobacco products are "dangerous" and "unsafe." Id., at 164-167. Thus, the FDA would apparently have to ban tobacco products, a result the court found clearly contrary to congressional intent. Ibid. This apparent anomaly, the Court of Appeals concluded, demonstrates that Congress did not intend to give the FDA authority to regulate tobacco. Id., at 167. The court also found that evidence external to the FDCA confirms this conclusion. Importantly, the FDA consistently stated before 1995 that it lacked jurisdiction over tobacco, and Congress has enacted [131] several tobacco-specific statutes fully cognizant of the FDA's position. See id., at 168-176. In fact, the court reasoned, Congress has considered and rejected many bills that would have given the agency such authority. See id., at 170-171. This, along with the absence of any intent by the enacting Congress in 1938 to subject tobacco products to regulation under the FDCA, demonstrates that Congress intended to withhold such authority from the FDA. Id., at 167-176. Having resolved the jurisdictional question against the agency, the Court of Appeals did not address whether the regulations exceed the FDA's authority under 21 U. S. C. § 360j(e) or violate the First Amendment. See 153 F. 3d, at 176, n. 29.

We granted the federal parties' petition for certiorari, 526 U. S. 1086 (1999), to determine whether the FDA has authority under the FDCA to regulate tobacco products as customarily marketed.

The FDA's assertion of jurisdiction to regulate tobacco products is founded on its conclusions that nicotine is a "drug" and that cigarettes and smokeless tobacco are "drug delivery devices." Again, the FDA found that tobacco products are "intended" to deliver the pharmacological effects of satisfying addiction, stimulation and tranquilization, and weight control because those effects are foreseeable to any reasonable manufacturer, consumers use tobacco products to obtain those effects, and tobacco manufacturers have designed their products to produce those effects. 61 Fed. Reg. 44632-44633 (1996). As an initial matter, respondents take issue with the FDA's reading of "intended," arguing that it is a term of art that refers exclusively to claims made by the manufacturer or vendor about the product. See Brief for Respondent Brown & Williamson Tobacco Corp. 6. That is, a product is not a drug or device under the FDCA unless the manufacturer or vendor makes some express claim concerning the product's therapeutic benefits. See id., at 6-7. We [132] need not resolve this question, however, because assuming, arguendo, that a product can be "intended to affect the structure or any function of the body" absent claims of therapeutic or medical benefit, the FDA's claim to jurisdiction contravenes the clear intent of Congress.

A threshold issue is the appropriate framework for analyzing the FDA's assertion of authority to regulate tobacco products. Because this case involves an administrative agency's construction of a statute that it administers, our analysis is governed by Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837 (1984). Under Chevron, a reviewing court must first ask "whether Congress has directly spoken to the precise question at issue." Id., at 842. IfCongress has done so, the inquiry is at an end; the court "must give effect to the unambiguously expressed intent of Congress." Id., at 843; see also United States v. Haggar Apparel Co., 526 U. S. 380, 392 (1999); Holly Farms Corp. v. NLRB, 517 U. S. 392, 398 (1996). But if Congress has not specifically addressed the question, a reviewing court must respect the agency's construction of the statute so long as it is permissible. See INS v. Aguirre-Aguirre, 526 U. S. 415, 424 (1999); Auer v. Robbins, 519 U. S. 452, 457 (1997). Such deference is justified because "[t]he responsibilities for assessing the wisdom of such policy choices and resolving the struggle between competing views of the public interest are not judicial ones," Chevron, supra, at 866, and because of the agency's greater familiarity with the everchanging facts and circumstances surrounding the subjects regulated, see Rust v. Sullivan, 500 U. S. 173, 187 (1991).

In determining whether Congress has specifically addressed the question at issue, a reviewing court should not confine itself to examining a particular statutory provision in isolation. The meaning—or ambiguity—of certain words or phrases may only become evident when placed in context. See Brown v. Gardner, 513 U. S. 115, 118 (1994) ("Ambiguity is a creature not of definitional possibilities but of statutory [133] context"). It is a "fundamental canon of statutory construction that the words of a statute must be read in their context and with a view to their place in the overall statutory scheme." Davis v. Michigan Dept. of Treasury, 489 U. S. 803, 809 (1989). A court must therefore interpret the statute "as a symmetrical and coherent regulatory scheme," Gustafson v. Alloyd Co., 513 U. S. 561, 569 (1995), and "fit, if possible, all parts into an harmonious whole," FTC v. Mandel Brothers, Inc., 359 U. S. 385, 389 (1959). Similarly, the meaning of one statute may be affected by other Acts, particularly where Congress has spoken subsequently and more specifically to the topic at hand. See United States v. Estate of Romani, 523 U. S. 517, 530-531 (1998); United States v. Fausto, 484 U. S. 439, 453 (1988). In addition, we must be guided to a degree by common sense as to the manner in which Congress is likely to delegate a policy decision of such economic and political magnitude to an administrative agency. Cf. MCI Telecommunications Corp. v. American Telephone & Telegraph Co., 512 U. S. 218, 231 (1994).

With these principles in mind, we find that Congress has directly spoken to the issue here and precluded the FDA's jurisdiction to regulate tobacco products.

Viewing the FDCA as a whole, it is evident that one of the Act's core objectives is to ensure that any product regulated by the FDA is "safe" and "effective" for its intended use. See 21 U. S. C. § 393(b)(2) (1994 ed., Supp. III) (defining the FDA's mission); More Information for Better Patient Care: Hearing before the Senate Committee on Labor and Human Resources, 104th Cong., 2d Sess., 83 (1996) (statement of FDA Deputy Comm'r Schultz) ("A fundamental precept of drug and device regulation in this country is that these products must be proven safe and effective before they can be sold"). This essential purpose pervades the FDCA. For instance, 21 U. S. C. § 393(b)(2) (1994 ed., Supp. III) defines [134] the FDA's "[m]ission" to include "protect[ing] the public health by ensuring that . . . drugsare safe and effective" and that "there is reasonable assurance of the safety and effectiveness of devices intended for human use." The FDCA requires premarket approval of any new drug, with some limited exceptions, and states that the FDA "shall issue an order refusing to approve the application" of a new drug if it is not safe and effective for its intended purpose. §§ 355(d)(1)-(2), (4)-(5). If the FDA discovers after approval that a drug is unsafe or ineffective, it "shall, after due notice and opportunity for hearing to the applicant, withdraw approval" of the drug. 21 U. S. C. §§ 355(e)(1)-(3). The Act also requires the FDA to classify all devices into one of three categories. § 360c(b)(1). Regardless of which category the FDA chooses, there must be a "reasonable assurance of the safety and effectiveness of the device." 21 U. S. C. §§ 360c(a)(1)(A)(i), (B), (C) (1994 ed. and Supp. III); 61 Fed. Reg. 44412 (1996). Even the "restricted device" provision pursuant to which the FDA promulgated the regulations at issue here authorizes the agency to place conditions on the sale or distribution of a device specifically when "there cannot otherwise be reasonable assurance of its safety and effectiveness." 21 U. S. C. § 360j(e). Thus, the Act generally requires the FDA to prevent the marketing of any drug or device where the "potential for inflicting death or physical injury is not offset by the possibility of therapeutic benefit." United States v. Rutherford, 442 U. S. 544, 556 (1979).

In its rulemaking proceeding, the FDA quite exhaustively documented that "tobacco products are unsafe," "dangerous," and "cause great pain and suffering from illness." 61 Fed. Reg. 44412 (1996). It found that the consumption of tobacco products presents "extraordinary health risks," and that "tobacco use is the single leading cause of preventable death in the United States." Id., at 44398. It stated that "[m]ore than 400,000 people die each year from tobaccorelated illnesses, such as cancer, respiratory illnesses, and [135] heart disease, often suffering long and painful deaths," and that "[t]obacco alone kills more people each year in the United States than acquired immunodeficiency syndrome (AIDS), car accidents, alcohol, homicides, illegal drugs, suicides, and fires, combined." Ibid. Indeed, the FDA characterized smoking as "a pediatric disease," id., at 44421, because "one out of every three young people who become regular smokers . . . will die prematurely as a result," id., at 44399.

These findings logically imply that, if tobacco products were "devices" under the FDCA, the FDA would be required to remove them from the market. Consider, first, the FDCA's provisions concerning the misbranding of drugs or devices. The Act prohibits "[t]he introduction or delivery for introduction into interstate commerce of any food, drug, device, or cosmetic that is adulterated or misbranded." 21 U. S. C. § 331(a). In light of the FDA's findings, two distinct FDCA provisions would render cigarettes and smokeless tobacco misbranded devices. First, § 352(j) deems a drug or device misbranded "[i]f it is dangerous to health when used in the dosage or manner, or with the frequency or duration prescribed, recommended, or suggested in the labeling thereof." The FDA's findings make clear that tobacco products are "dangerous to health" when used in the manner prescribed. Second, a drug or device is misbranded under the Act "[u]nless its labeling bears . . . adequate directions for use . . . in such manner and form, as are necessary for the protection of users," except where such directions are "not necessary for the protection of the public health." § 352(f)(1). Given the FDA's conclusions concerning the health consequences of tobacco use, there are no directions that could adequately protect consumers. That is, there are no directions that could make tobacco products safe for obtaining their intended effects. Thus, were tobacco products within the FDA's jurisdiction, the Act would deem them misbranded devices that could not be introduced into interstate [136] commerce. Contrary to the dissent's contention, the Act admits no remedial discretion once it is evident that the device is misbranded.

Second, the FDCA requires the FDA to place all devices that it regulates into one of three classifications. See § 360c(b)(1). The agency relies on a device's classification in determining the degree of control and regulation necessary to ensure that there is "a reasonable assurance of safety and effectiveness." 61 Fed. Reg. 44412 (1996). The FDA has yet to classify tobacco products. Instead, the regulations at issue here represent so-called "general controls," which the Act entitles the agency to impose in advance of classification. See id., at 44404-44405. Although the FDCA prescribes no deadline for device classification, the FDA has stated that it will classify tobacco products "in a future rulemaking" as required by the Act. Id., at 44412. Given the FDA's findings regarding the health consequences of tobacco use, the agency would have to place cigarettes and smokeless tobacco in Class III because, even after the application of the Act's available controls, they would "presen[t] a potential unreasonable risk of illness or injury." 21 U. S. C. § 360c(a)(1)(C). As Class III devices, tobacco products would be subject to the FDCA's premarket approval process. See 21 U. S. C. § 360c(a)(1)(C) (1994 ed., Supp. III); 21 U. S. C. § 360e; 61 Fed. Reg. 44412 (1996). Under these provisions, the FDA would be prohibited from approving an application for premarket approval without "a showing of reasonable assurance that such device is safe under the conditions of use prescribed, recommended, or suggested in the proposed labeling thereof." 21 U. S. C. § 360e(d)(2)(A). In view of the FDA's conclusions regarding the health effects of tobacco use, the agency would have no basis for finding any such reasonable assurance of safety. Thus, once the FDA fulfilled its statutory obligation to classify tobacco products, it could not allow them to be marketed.

[137] The FDCA's misbranding and device classification provisions therefore make evident that were the FDA to regulate cigarettes and smokeless tobacco, the Act would require the agency to ban them. In fact, based on these provisions, the FDA itself has previously taken the position that if tobacco products were within its jurisdiction, "they would have to be removed from the market because it would be impossible to prove they were safe for their intended us[e]." Public Health Cigarette Amendments of 1971: Hearings before the Commerce Subcommittee on S. 1454, 92d Cong., 2d Sess., 239 (1972) (hereinafter 1972 Hearings) (statement of FDA Comm'r Charles Edwards). See also Cigarette Labeling and Advertising: Hearings before the House Committee on Interstate and Foreign Commerce, 88th Cong., 2d Sess., 18 (1964) (hereinafter 1964 Hearings) (statement of Dept. of Health, Education, and Welfare (HEW) Secretary Anthony Celebrezze that proposed amendments to the FDCA that would have given the FDA jurisdiction over "smoking product[s]" "might well completely outlaw at least cigarettes").

Congress, however, has foreclosed the removal of tobacco products from the market. A provision of the United States Code currently in force states that "[t]he marketing of tobacco constitutes one of the greatest basic industries of the United States with ramifying activities which directly affect interstate and foreign commerce at every point, and stable conditions therein are necessary to the general welfare." 7 U. S. C. § 1311(a). More importantly, Congress has directly addressed the problem of tobacco and health through legislation on six occasions since 1965. See Federal Cigarette Labeling and Advertising Act (FCLAA), Pub. L. 89-92, 79 Stat. 282; Public Health Cigarette Smoking Act of 1969, Pub. L. 91-222, 84 Stat. 87; Alcohol and Drug Abuse Amendments of 1983, Pub. L. 98-24, 97 Stat. 175; Comprehensive Smoking Education Act, Pub. L. 98-474, 98 Stat. 2200; Comprehensive Smokeless Tobacco Health Education Act of 1986, Pub. L. 99-252, 100 Stat. 30; Alcohol, Drug Abuse, and Mental [138] Health Administration Reorganization Act, Pub. L. 102-321, § 202, 106 Stat. 394. When Congress enacted these statutes, the adverse health consequences of tobacco use were well known, as were nicotine's pharmacological effects. See, e. g., U. S. Dept. of Health, Education, and Welfare, U. S. Surgeon General's Advisory Committee, Smoking and Health 25-40, 69-75 (1964) (hereinafter 1964 Surgeon General's Report) (concluding that cigarette smoking causes lung cancer, coronary artery disease, and chronic bronchitis and emphysema, and that nicotine has various pharmacological effects, including stimulation, tranquilization, and appetite suppression); U. S. Dept. of Health and Human Services, Public Health Service, Health Consequences of Smoking for Women 7-12 (1980) (finding that mortality rates for lung cancer, chronic lung disease, and coronary heart disease are increased for both women and men smokers, and that smoking during pregnancy is associated with significant adverse health effects on the unborn fetus and newborn child); U. S. Dept. of Health and Human Services, Public Health Service, Why People Smoke Cigarettes (1983), in Smoking Prevention Education Act, Hearings on H. R. 1824 before the Subcommittee on Health and the Environment of the House Committee on Energy and Commerce, 98th Cong., 1st Sess., 32-37 (1983) (hereinafter 1983 House Hearings) (stating that smoking is "the most widespread example of drug dependence in our country," and that cigarettes "affect the chemistry of the brain and nervous system"); U. S. Dept. of Health and Human Services, Public Health Service, The Health Consequences of Smoking: Nicotine Addiction 6-9, 145-239 (1988) (hereinafter 1988 Surgeon General's Report) (concluding that tobacco products are addicting in much the same way as heroin and cocaine, and that nicotine is the drug that causes addiction). Nonetheless, Congress stopped well short of ordering a ban. Instead, it has generally regulated the labeling and advertisement of tobacco products, expressly providing that it is the policy of Congress that "commerce and the national [139] economy may be . . . protected to the maximum extent consistent with" consumers "be[ing] adequately informed about any adverse health effects." 15 U. S. C. § 1331. Congress' decisions to regulate labeling and advertising and to adopt the express policy of protecting "commerce and the national economy . . . to the maximum extent" reveal its intent that tobacco products remain on the market. Indeed, the collective premise of these statutes is that cigarettes and smokeless tobacco will continue to be sold in the United States. A ban of tobacco products by the FDA would therefore plainly contradict congressional policy.

The FDA apparently recognized this dilemma and concluded, somewhat ironically, that tobacco products are actually "safe" within the meaning of the FDCA. In promulgating its regulations, the agency conceded that "tobacco products are unsafe, as that term is conventionally understood." 61 Fed. Reg. 44412 (1996). Nonetheless, the FDA reasoned that, in determining whether a device is safe under the Act, it must consider "not only the risks presented by a product but also any of the countervailing effects of use of that product, including the consequences of not permitting the product to be marketed." Id., at 44412-44413. Applying this standard, the FDA found that, because of the high level of addiction among tobacco users, a ban would likely be "dangerous." Id., at 44413. In particular, current tobacco users could suffer from extreme withdrawal, the health care system and available pharmaceuticals might not be able to meet the treatment demands of those suffering from withdrawal, and a black market offering cigarettes even more dangerous than those currently sold legally would likely develop. Ibid. The FDA therefore concluded that, "while taking cigarettes and smokeless tobacco off the market could prevent some people from becoming addicted and reduce death and disease for others, the record does not establish that such a ban is the appropriate public health response under the act." Id., at 44398.

[140] It may well be, as the FDA asserts, that "these factors must be considered when developing a regulatory scheme that achieves the best public health result for these products." Id., at 44413. But the FDA's judgment that leaving tobacco products on the market "is more effective in achieving public health goals than a ban," ibid., is no substitute for the specific safety determinations required by the FDCA's various operative provisions. Several provisions in the Act require the FDA to determine that the product itself is safe as used by consumers. That is, the product's probable therapeutic benefits must outweigh its risk of harm. See United States v. Rutherford, 442 U. S., at 555 ("[T]he Commissioner generally considers a drug safe when the expected therapeutic gain justifies the risk entailed by its use"). In contrast, the FDA's conception of safety would allow the agency, with respect to each provision of the FDCA that requires the agency to determine a product's "safety" or "dangerousness," to compare the aggregate health effects of alternative administrative actions. This is a qualitatively different inquiry. Thus, although the FDA has concluded that a ban would be "dangerous," it has not concluded that tobacco products are "safe" as that term is used throughout the Act.

Consider 21 U. S. C. § 360c(a)(2), which specifies those factors that the FDA may consider in determining the safety and effectiveness of a device for purposes of classification, performance standards, and premarket approval. For all devices regulated by the FDA, there must at least be a "reasonable assurance of the safety and effectiveness of the device." See 21 U. S. C. §§ 360c(a)(1)(A)(i), (B), (C) (1994 ed. and Supp. III); 61 Fed. Reg. 44412 (1996). Title 21 U. S. C. § 360c(a)(2) provides that

"the safety and effectiveness of a device are to be determined—

"(A) with respect to the persons for whose use the device is represented or intended,

[141] "(B) with respect to the conditions of use prescribed, recommended, or suggested in the labeling of the device, and

"(C) weighing any probable benefit to health from the use of the device against any probable risk of injury or illness from such use."

A straightforward reading of this provision dictates that the FDA must weigh the probable therapeutic benefits of the device to the consumer against the probable risk of injury. Applied to tobacco products, the inquiry is whether their purported benefits—satisfying addiction, stimulation and sedation, and weight control—outweigh the risks to health from their use. To accommodate the FDA's conception of safety, however, one must read "any probable benefit to health" to include the benefit to public health stemming from adult consumers' continued use of tobacco products, even though the reduction of tobacco use is the raison d'être of the regulations. In other words, the FDA is forced to contend that the very evil it seeks to combat is a "benefit to health." This is implausible.

The FDA's conception of safety is also incompatible with the FDCA's misbranding provision. Again, § 352(j) provides that a product is "misbranded" if "it is dangerous to health when used in the dosage or manner, or with the frequency or duration prescribed, recommended, or suggested in the labeling thereof." According to the FDA's understanding, a product would be "dangerous to health," and therefore misbranded under § 352(j), when, in comparison to leaving the product on the market, a ban would not produce "adverse health consequences" in aggregate. Quite simply, these are different inquiries. Although banning a particular product might be detrimental to public health in aggregate, the product could still be "dangerous to health" when used as directed. Section 352(j) focuses on dangers to the consumer from use of the product, not those stemming from the agency's remedial measures.

[142] Consequently, the analogy made by the FDA and the dissent to highly toxic drugs used in the treatment of various cancers is unpersuasive. See 61 Fed. Reg. 44413 (1996); post, at 177 (opinion of Breyer, J.). Although "dangerous" in some sense, these drugs are safe within the meaning of the Act because, for certain patients, the therapeutic benefits outweigh the risk of harm. Accordingly, such drugs cannot properly be described as "dangerous to health" under 21 U. S. C. § 352(j). The same is not true for tobacco products. As the FDA has documented in great detail, cigarettes and smokeless tobacco are an unsafe means to obtaining any pharmacological effect.

The dissent contends that our conclusion means that "the FDCA requires the FDA to ban outright `dangerous' drugs or devices," post, at 174, and that this is a "perverse" reading of the statute, post, at 174, 180. This misunderstands our holding. The FDA, consistent with the FDCA, may clearly regulate many "dangerous" products without banning them. Indeed, virtually every drug or device poses dangers under certain conditions. What the FDA may not do is conclude that a drug or device cannot be used safely for any therapeutic purpose and yet, at the same time, allow that product to remain on the market. Such regulation is incompatible with the FDCA's core objective of ensuring that every drug or device is safe and effective.

Considering the FDCA as a whole, it is clear that Congress intended to exclude tobacco products from the FDA's jurisdiction. A fundamental precept of the FDCA is that any product regulated by the FDA—but not banned—must be safe for its intended use. Various provisions of the Act make clear that this refers to the safety of using the product to obtain its intended effects, not the public health ramifications of alternative administrative actions by the FDA. That is, the FDA must determine that there is a reasonable assurance that the product's therapeutic benefits outweigh the risk of harm to the consumer. According to this standard, [143] the FDA has concluded that, although tobacco products might be effective in delivering certain pharmacological effects, they are "unsafe" and "dangerous" when used for these purposes. Consequently, if tobacco products were within the FDA's jurisdiction, the Act would require the FDA to remove them from the market entirely. But a ban would contradict Congress' clear intent as expressed in its more recent, tobacco-specific legislation. The inescapable conclusion is that there is no room for tobacco products within the FDCA's regulatory scheme. If they cannot be used safely for any therapeutic purpose, and yet they cannot be banned, they simply do not fit.

In determining whether Congress has spoken directly to the FDA's authority to regulate tobacco, we must also consider in greater detail the tobacco-specific legislation that Congress has enacted over the past 35 years. At the time a statute is enacted, it may have a range of plausible meanings. Over time, however, subsequent acts can shape or focus those meanings. The "classic judicial task of reconciling many laws enacted over time, and getting them to `make sense' in combination, necessarily assumes that the implications of a statute may be altered by the implications of a later statute." United States v. Fausto, 484 U. S., at 453. This is particularly so where the scope of the earlier statute is broad but the subsequent statutes more specifically address the topic at hand. As we recognized recently in United States v. Estate of Romani, "a specific policy embodied in a later federal statute should control our construction of the [earlier] statute, even though it ha[s] not been expressly amended." 523 U. S., at 530-531.

Congress has enacted six separate pieces of legislation since 1965 addressing the problem of tobacco use and human health. See supra, at 137-138. Those statutes, among other things, require that health warnings appear on all packaging and in all print and outdoor advertisements, see [144] 15 U. S. C. §§ 1331, 1333, 4402; prohibit the advertisement of tobacco products through "any medium of electronic communication" subject to regulation by the Federal Communications Commission (FCC), see §§ 1335, 4402(f); require the Secretary of HHS to report every three years to Congress on research findings concerning "the addictive property of tobacco," 42 U. S. C. § 290aa—2(b)(2); and make States' receipt of certain federal block grants contingent on their making it unlawful "for any manufacturer, retailer, or distributor of tobacco products to sell or distribute any such product to any individual under the age of 18," § 300x—26(a)(1).

In adopting each statute, Congress has acted against the backdrop of the FDA's consistent and repeated statements that it lacked authority under the FDCA to regulate tobacco absent claims of therapeutic benefit by the manufacturer. In fact, on several occasions over this period, and after the health consequences of tobacco use and nicotine's pharmacological effects had become well known, Congress considered and rejected bills that would have granted the FDA such jurisdiction. Under these circumstances, it is evident that Congress' tobacco-specific statutes have effectively ratified the FDA's long-held position that it lacks jurisdiction under the FDCA to regulate tobacco products. Congress has created a distinct regulatory scheme to address the problem of tobacco and health, and that scheme, as presently constructed, precludes any role for the FDA.

On January 11, 1964, the Surgeon General released the report of the Advisory Committee on Smoking and Health. That report documented the deleterious health effects of smoking in great detail, concluding, in relevant part, "that cigarette smoking contributes substantially to mortality from certain specific diseases and to the overall death rate." 1964 Surgeon General's Report 31. It also identified the pharmacological effects of nicotine, including "stimulation," "tranquilization," and "suppression of appetite." Id., at 74-75. Seven days after the report's release, the Federal Trade [145] Commission (FTC) issued a notice of proposed rulemaking, see 29 Fed. Reg. 530-532 (1964), and in June 1964, the FTC promulgated a final rule requiring cigarette manufacturers "to disclose, clearly and prominently, in all advertising and on every pack, box, carton or other container . . . that cigarette smoking is dangerous to health and may cause death from cancer and other diseases," id., at 8325. The rule was to become effective January 1, 1965, but, on a request from Congress, the FTC postponed enforcement for six months. See Cipollone v. Liggett Group, Inc., 505 U. S. 504, 513-514 (1992).

In response to the Surgeon General's report and the FTC's proposed rule, Congress convened hearings to consider legislation addressing "the tobacco problem." 1964 Hearings 1. During those deliberations, FDA representatives testified before Congress that the agency lacked jurisdiction under the FDCA to regulate tobacco products. Surgeon General Terry was asked during hearings in 1964 whether HEW had the "authority to brand or label the packages of cigarettes or to control the advertising there." Id., at 56. The Surgeon General stated that "we do not have such authority in existing laws governing the . . . Food and Drug Administration." Ibid. Similarly, FDA Deputy Commissioner Rankin testified in 1965 that "[t]he Food and Drug Administration has no jurisdiction under the Food, Drug, and Cosmetic Act over tobacco, unless it bears drug claims." Cigarette Labeling and Advertising—1965: Hearings on H. R. 2248 before the House Committee on Interstate and Foreign Commerce, 89th Cong., 1st Sess., 193 (hereinafter 1965 Hearings). See also Letter to Directors of Bureaus, Divisions and Directors of Districts from FDA Bureau of Enforcement (May 24, 1963), in 1972 Hearings 240 ("[T]obacco marketed for chewing or smoking without accompanying therapeutic claims, does not meet the definitions in the Food, Drug, and Cosmetic Act for food, drug, device or cosmetic"). In fact, HEW Secretary Celebrezze urged Congress not to amend the FDCA to cover [146] "smoking products" because, in light of the findings in the Surgeon General's report, such a "provision might well completely outlaw at least cigarettes. This would be contrary to what, we understand, is intended or what, in the light of our experience with the 18th amendment, would be acceptable to the American people." 1964 Hearings 18.

The FDA's disavowal of jurisdiction was consistent with the position that it had taken since the agency's inception. As the FDA concedes, it never asserted authority to regulate tobacco products as customarily marketed until it promulgated the regulations at issue here. See Brief for Petitioners 37; see also Brief for Appellee (FDA) in Action on Smoking and Health v. Harris, 655 F. 2d 236 (CADC 1980), in 9 Rec. in No. 97-1604 (CA4), Tab No. 4, pp. 14-15 ("In the 73 years since the enactment of the original Food and Drug Act, and in the 41 years since the promulgation of the modern Food, Drug, and Cosmetic Act, the FDA has repeatedly informed Congress that cigarettes are beyond the scope of the statute absent health claims establishing a therapeutic intent on behalf of the manufacturer or vendor").

The FDA's position was also consistent with Congress' specific intent when it enacted the FDCA. Before the Act's adoption in 1938, the FDA's predecessor agency, the Bureau of Chemistry, announced that it lacked authority to regulate tobacco products under the Pure Food and Drug Act of 1906, ch. 3915, 34 Stat. 768, unless they were marketed with therapeutic claims. See U. S. Dept. of Agriculture, Bureau of Chemistry, 13 Service and Regulatory Announcements 24 (Apr. 1914) (Feb. 1914 Announcements ¶ 13, Opinion of Chief of Bureau C. L. Alsberg). In 1929, Congress considered and rejected a bill "[t]o amend the Food and Drugs Act of June 30, 1906, by extending its provisions to tobacco and tobacco products." S. 1468, 71st Cong., 1st Sess., 1. See also 71 Cong. Rec. 2589 (1929) (remarks of Sen. Smoot). And, as the FDA admits, there is no evidence in the text of the FDCA or its legislative history that Congress in 1938 even considered [147] the applicability of the Act to tobacco products. See Brief for Petitioners 22, n. 4. Given the economic and political significance of the tobacco industry at the time, it is extremely unlikely that Congress could have intended to place tobacco within the ambit of the FDCA absent any discussion of the matter. Of course, whether the Congress that enacted the FDCA specifically intended the Act to cover tobacco products is not determinative; "it is ultimately the provisions of our laws rather than the principal concerns of our legislators by which we are governed." Oncale v. Sundowner Offshore Services, Inc., 523 U. S. 75, 79 (1998); see also TVA v. Hill, 437 U. S. 153, 185 (1978) ("It is not for us to speculate, much less act, on whether Congress would have altered its stance had the specific events of this case been anticipated"). Nonetheless, this intent is certainly relevant to understanding the basis for the FDA's representations to Congress and the background against which Congress enacted subsequent tobacco-specific legislation.

Moreover, before enacting the FCLAA in 1965, Congress considered and rejected several proposals to give the FDA the authority to regulate tobacco. In April 1963, Representative Udall introduced a bill "[t]o amend the Federal Food, Drug, and Cosmetic Act so as to make that Act applicable to smoking products." H. R. 5973, 88th Cong., 1st Sess., 1. Two months later, Senator Moss introduced an identical bill in the Senate. S. 1682, 88th Cong., 1st Sess. (1963). In discussing his proposal on the Senate floor, Senator Moss explained that "this amendment simply places smoking products under FDA jurisdiction, along with foods, drugs, and cosmetics." 109 Cong. Rec. 10322 (1963). In December 1963, Representative Rhodes introduced another bill that would have amended the FDCA "by striking out `food, drug, device, or cosmetic, each place where it appears therein and inserting in lieu thereof `food, drug, device, cosmetic, or smoking product.' " H. R. 9512, 88th Cong., 1st Sess., § 3 (1963). And in January 1965, five months before passage of [148] the FCLAA, Representative Udall again introduced a bill to amend the FDCA "to make that Act applicable to smoking products." H. R. 2248, 89th Cong., 1st Sess., 1. None of these proposals became law.

Congress ultimately decided in 1965 to subject tobacco products to the less extensive regulatory scheme of the FCLAA, which created a "comprehensive Federal program to deal with cigarette labeling and advertising with respect to any relationship between smoking and health." Pub. L. 89-92, § 2, 79 Stat. 282. The FCLAA rejected any regulation of advertising, but it required the warning, "Caution: Cigarette Smoking May Be Hazardous to Your Health," to appear on all cigarette packages. Id., § 4, 79 Stat. 283. In the FCLAA's "Declaration of Policy," Congress stated that its objective was to balance the goals of ensuring that "the public may be adequately informed that cigarette smoking may be hazardous to health" and protecting "commerce and the national economy . . . to the maximum extent." Id., § 2, 79 Stat. 282 (codified at 15 U. S. C. § 1331).

Not only did Congress reject the proposals to grant the FDA jurisdiction, but it explicitly pre-empted any other regulation of cigarette labeling: "No statement relating to smoking and health, other than the statement required by . . . this Act, shall be required on any cigarette package." Pub. L. 89-92, § 5(a), 79 Stat. 283. The regulation of product labeling, however, is an integral aspect of the FDCA, both as it existed in 1965 and today. The labeling requirements currently imposed by the FDCA, which are essentially identical to those in force in 1965, require the FDA to regulate the labeling of drugs and devices to protect the safety of consumers. See 21 U. S. C. § 352; 21 U. S. C. § 352 (1964 ed. and Supp. IV). As discussed earlier, the Act requires that all products bear "adequate directions for use . . . as are necessary for the protection of users," 21 U. S. C. § 352(f)(1); 21 U. S. C. § 352(f)(1) (1964 ed.); requires that all products provide "adequate warnings against use in those pathological [149] conditions or by children where its use may be dangerous to health," 21 U. S. C. § 352(f)(2); 21 U. S. C. § 352(f)(2) (1964 ed.); and deems a product misbranded "[i]f it is dangerous to health when used in the dosage or manner, or with the frequency or duration prescribed, recommended, or suggested in the labeling thereof," 21 U. S. C. § 352(j); 21 U. S. C. § 352(j) (1964 ed.). In this sense, the FCLAA was—and remains—incompatible with FDA regulation of tobacco products. This is not to say that the FCLAA's pre-emption provision by itself necessarily foreclosed FDA jurisdiction. See Cipollone v. Liggett Group, Inc., 505 U. S., at 518-519. But it is an important factor in assessing whether Congress ratified the agency's position—that is, whether Congress adopted a regulatory approach to the problem of tobacco and health that contemplated no role for the FDA.

Further, the FCLAA evidences Congress' intent to preclude any administrative agency from exercising significant policymaking authority on the subject of smoking and health. In addition to prohibiting any additional requirements for cigarette labeling, the FCLAA provided that "[n]o statement relating to smoking and health shall be required in the advertising of any cigarettes the packages of which are labeled in conformity with the provisions of this Act." Pub. L. 89-92, § 5(b), 79 Stat. 283. Thus, in reaction to the FTC's attempt to regulate cigarette labeling and advertising, Congress enacted a statute reserving exclusive control over both subjects to itself.

Subsequent tobacco-specific legislation followed a similar pattern. By the FCLAA's own terms, the prohibition on any additional cigarette labeling or advertising regulations relating to smoking and health was to expire July 1, 1969. See § 10, 79 Stat. 284. In anticipation of the provision's expiration, both the FCC and the FTC proposed rules governing the advertisement of cigarettes. See 34 Fed. Reg. 1959 (1969) (FCC proposed rule to "ban the broadcast of cigarette commercials by radio and television stations"); id., at 7917 [150] (FTC proposed rule requiring manufacturers to disclose on all packaging and in all print advertising "`that cigarette smoking is dangerous to health and may cause death from cancer, coronary heart disease, chronic bronchitis, pulmonary emphysema, and other diseases' "). After debating the proper role for administrative agencies in the regulation of tobacco, see generally Cigarette Labeling and Advertising— 1969: Hearings before the House Committee on Interstate and Foreign Commerce, 91st Cong., 1st Sess., pt. 2 (1969), Congress amended the FCLAA by banning cigarette advertisements "on any medium of electronic communication subject to the jurisdiction of the Federal Communications Commission" and strengthening the warning required to appear on cigarette packages. Public Health Cigarette Smoking Act of 1969, Pub. L. 91-222, §§ 4, 6, 84 Stat. 88-89. Importantly, Congress extended indefinitely the prohibition on any other regulation of cigarette labeling with respect to smoking and health (again despite the importance of labeling regulation under the FDCA). § 5(a), 84 Stat. 88 (codified at 15 U. S. C. § 1334(a)). Moreover, it expressly forbade the FTC from taking any action on its pending rule until July 1, 1971, and it required the FTC, if it decided to proceed with its rule thereafter, to notify Congress at least six months in advance of the rule's becoming effective. § 7(a), 84 Stat. 89. As the chairman of the House committee in which the bill originated stated, "the Congress—the body elected by the people— must make the policy determinations involved in this legislation—and not some agency made up of appointed officials." 116 Cong. Rec. 7920 (1970) (remarks of Rep. Staggers).

Four years later, after Congress had transferred the authority to regulate substances covered by the Hazardous Substances Act (HSA) from the FDA to the Consumer Products Safety Commission (CPSC), the American Public Health Association, joined by Senator Moss, petitioned the CPSC to regulate cigarettes yielding more than 21 milligrams of tar. See Action on Smoking and Health v. Harris, 655 F. 2d 236, [151] 241 (CADC 1980); R. Kluger, Ashes to Ashes 375-376 (1996). After the CPSC determined that it lacked authority under the HSA to regulate cigarettes, a District Court held that the HSA did, in fact, grant the CPSC such jurisdiction and ordered it to reexamine the petition. See American Public Health Association v. Consumer Product Safety Commission, [1972-1975 Transfer Binder] CCH Consumer Prod. Safety Guide ¶ 75,081 (DC 1975), vacated as moot, No. 75-1863 (CADC 1976). Before the CPSC could take any action, however, Congress mooted the issue by adopting legislation that eliminated the agency's authority to regulate "tobacco and tobacco products." Consumer Product Safety Commission Improvements Act of 1976, Pub. L. 94-284, § 3(c), 90 Stat. 503 (codified at 15 U. S. C. § 1261(f)(2)). Senator Moss acknowledged that the "legislation, in effect, reverse[d]" the District Court's decision, 121 Cong. Rec. 23563 (1975), and the FDA later observed that the episode was "particularly" "indicative of the policy of Congress to limit the regulatory authority over cigarettes by Federal Agencies," Letter to Action on Smoking and Health (ASH) Executive Director Banzhaf from FDA Comm'r Goyan (Nov. 25, 1980), App. 59. A separate statement in the Senate Report underscored that the legislation's purpose was to "unmistakably reaffirm the clear mandate of the Congress that the basic regulation of tobacco and tobacco products is governed by the legislation dealing with the subject, . . . and that any further regulation in this sensitive and complex area must be reserved for specific Congressional action." S. Rep. No. 94-251, p. 43 (1975) (additional views of Sens. Hartke, Hollings, Ford, Stevens, and Beall).

Meanwhile, the FDA continued to maintain that it lacked jurisdiction under the FDCA to regulate tobacco products as customarily marketed. In 1972, FDA Commissioner Edwards testified before Congress that "cigarettes recommended for smoking pleasure are beyond the Federal Food, Drug, and Cosmetic Act." 1972 Hearings 239, 242. He further [152] stated that the FDA believed that the Public Health Cigarette Smoking Act "demonstrates that the regulation of cigarettes is to be the domain of Congress," and that "labeling or banning cigarettes is a step that can be take[n] only by the Congress. Any such move by FDA would be inconsistent with the clear congressional intent." Ibid.

In 1977, ASH filed a citizen petition requesting that the FDA regulate cigarettes, citing many of the same grounds that motivated the FDA's rulemaking here. See Citizen Petition, No. 77P-0185 (May 26, 1977), 10 Rec. in No. 97-1604 (CA4), Tab No. 22, pp. 1-10. ASH asserted that nicotine was highly addictive and had strong physiological effects on the body; that those effects were "intended" because consumers use tobacco products precisely to obtain those effects; and that tobacco causes thousands of premature deaths annually. Ibid. In denying ASH's petition, FDA Commissioner Kennedy stated that "[t]he interpretation of the Act by FDA consistently has been that cigarettes are not a drug unless health claims are made by the vendors." Letter to ASH Executive Director Banzhaf (Dec. 5, 1977), App. 47. After the matter proceeded to litigation, the FDA argued in its brief to the Court of Appeals that "cigarettes are not comprehended within the statutory definition of the term `drug' absent objective evidence that vendors represent or intend that their products be used as a drug." Brief for Appellee in Action on Smoking and Health v. Harris, 655 F. 2d 236 (CADC 1980), 9 Rec. in No. 97-1604 (CA4), Tab No. 4, at 27-28. The FDA also contended that Congress had "long been aware that the FDA does not consider cigarettes to be within its regulatory authority in the absence of health claims made on behalf of the manufacturer or vendor," and that, because "Congress has never acted to disturb the agency's interpretation," it had "acquiesced in the FDA's interpretation of the statutory limits on its authority to regulate cigarettes." Id., at 23, 27, n. 23. The Court of Appeals upheld the FDA's position, concluding that "[i]f the statute [153] requires expansion, that is the job of Congress." Action on Smoking and Health v. Harris, 655 F. 2d, at 243. In 1980, the FDA also denied a request by ASH to commence rulemaking proceedings to establish the agency's jurisdiction to regulate cigarettes as devices. See Letter to ASH Executive Director Banzhaf from FDA Comm'r Goyan (Nov. 25, 1980), App. 50-51. The agency stated that "[i]nsofar as rulemaking would relate to cigarettes or attached filters as customarily marketed, we have concluded that FDA has no jurisdiction under section 201(h) of the Act [21 U. S. C. § 321(h)]." Id., at 67.

In 1983, Congress again considered legislation on the subject of smoking and health. HHS Assistant Secretary Brandt testified that, in addition to being "a major cause of cancer," smoking is a "major cause of heart disease" and other serious illnesses, and can result in "unfavorable pregnancy outcomes." 1983 House Hearings 19-20. He also stated that it was "well-established that cigarette smoking is a drug dependence, and that smoking is addictive for many people." Id., at 20. Nonetheless, Assistant Secretary Brandt maintained that "the issue of regulation of tobacco . . . is something that Congress has reserved to itself, and we do not within the Department have the authority to regulate nor are we seeking such authority." Id., at 74. He also testified before the Senate, stating that, despite the evidence of tobacco's health effects and addictiveness, the Department's view was that "Congress has assumed the responsibility of regulating . . . cigarettes." Smoking Prevention and Education Act: Hearings on S. 772 before the Senate Committee on Labor and Human Resources, 98th Cong., 1st Sess., 56 (1983) (hereinafter 1983 Senate Hearings).

Against this backdrop, Congress enacted three additional tobacco-specific statutes over the next four years that incrementally expanded its regulatory scheme for tobacco products. In 1983, Congress adopted the Alcohol and Drug Abuse Amendments, Pub. L. 98-24, 97 Stat. 175 (codified at [154] 42 U. S. C. § 290aa et seq. ), which require the Secretary of HHS to report to Congress every three years on the "addictive property of tobacco" and to include recommendations for action that the Secretary may deem appropriate. A year later, Congress enacted the Comprehensive Smoking Education Act, Pub. L. 98-474, 98 Stat. 2200, which amended the FCLAA by again modifying the prescribed warning. Notably, during debate on the Senate floor, Senator Hawkins argued that the FCLAA was necessary in part because "[u]nder the Food, Drug and Cosmetic Act, the Congress exempted tobacco products." 130 Cong. Rec. 26953 (1984). And in 1986, Congress enacted the Comprehensive Smokeless Tobacco Health Education Act of 1986 (CSTHEA), Pub. L. 99-252, 100 Stat. 30 (codified at 15 U. S. C. § 4401 et seq. ), which essentially extended the regulatory provisions of the FCLAA to smokeless tobacco products. Like the FCLAA, the CSTHEA provided that "[n]o statement relating to the use of smokeless tobacco products and health, other than the statements required by [the Act], shall be required by any Federal agency to appear on any package . . . of a smokeless tobacco product." § 7(a), 100 Stat. 34 (codified at 15 U. S. C. § 4406(a)). Thus, as with cigarettes, Congress reserved for itself an aspect of smokeless tobacco regulation that is particularly important to the FDCA's regulatory scheme.

In 1988, the Surgeon General released a report summarizing the abundant scientific literature demonstrating that "[c]igarettes and other forms of tobacco are addicting," and that "nicotine is psychoactive" and "causes physical dependence characterized by a withdrawal syndrome that usually accompanies nicotine abstinence." 1988 Surgeon General's Report 14. The report further concluded that the "pharmacologic and behavioral processes that determine tobacco addiction are similar to those that determine addiction to drugs such as heroin and cocaine." Id., at 15. In the same year, FDA Commissioner Young stated before Congress that "it doesn't look like it is possible to regulate [tobacco] under the [155] Food, Drug and Cosmetic Act even though smoking, I think, has been widely recognized as being harmful to human health." Rural Development, Agriculture, and Related Agencies Appropriations for 1989: Hearings before a Subcommittee of the House Committee on Appropriations, 100th Cong., 2d Sess., 409 (1988). At the same hearing, the FDA's General Counsel testified that "what is fairly important in FDA law is whether a product has a therapeutic purpose," and "[c]igarettes themselves are not used for a therapeutic purpose as that concept is ordinarily understood." Id., at 410. Between 1987 and 1989, Congress considered three more bills that would have amended the FDCA to grant the FDA jurisdiction to regulate tobacco products. See H. R. 3294, 100th Cong., 1st Sess. (1987); H. R. 1494, 101st Cong., 1st Sess. (1989); S. 769, 101st Cong., 1st Sess. (1989). As before, Congress rejected the proposals. In 1992, Congress instead adopted the Alcohol, Drug Abuse, and Mental Health Administration Reorganization Act, Pub. L. 102-321, § 202, 106 Stat. 394 (codified at 42 U. S. C. § 300x et seq. ), which creates incentives for States to regulate the retail sale of tobacco products by making States' receipt of certain block grants contingent on their prohibiting the sale of tobacco products to minors.

Taken together, these actions by Congress over the past 35 years preclude an interpretation of the FDCA that grants the FDA jurisdiction to regulate tobacco products. We do not rely on Congress' failure to act—its consideration and rejection of bills that would have given the FDA this authority—in reaching this conclusion. Indeed, this is not a case of simple inaction by Congress that purportedly represents its acquiescence in an agency's position. To the contrary, Congress has enacted several statutes addressing the particular subject of tobacco and health, creating a distinct regulatory scheme for cigarettes and smokeless tobacco. In doing so, Congress has been aware of tobacco's health hazards and its pharmacological effects. It has also enacted this legislation [156] against the background of the FDA repeatedly and consistently asserting that it lacks jurisdiction under the FDCA to regulate tobacco products as customarily marketed. Further, Congress has persistently acted to preclude a meaningful role for any administrative agency in making policy on the subject of tobacco and health. Moreover, the substance of Congress' regulatory scheme is, in an important respect, incompatible with FDA jurisdiction. Although the supervision of product labeling to protect consumer health is a substantial component of the FDA's regulation of drugs and devices, see 21 U. S. C. § 352 (1994 ed. and Supp. III), the FCLAA and the CSTHEA explicitly prohibit any federal agency from imposing any health-related labeling requirements on cigarettes or smokeless tobacco products, see 15 U. S. C. §§ 1334(a), 4406(a).

Under these circumstances, it is clear that Congress' tobacco-specific legislation has effectively ratified the FDA's previous position that it lacks jurisdiction to regulate tobacco. As in Bob Jones Univ. v. United States, 461 U. S. 574 (1983), "[i]t is hardly conceivable that Congress—and in this setting, any Member of Congress—was not abundantly aware of what was going on." Id., at 600-601. Congress has affirmatively acted to address the issue of tobacco and health, relying on the representations of the FDA that it had no authority to regulate tobacco. It has created a distinct scheme to regulate the sale of tobacco products, focused on labeling and advertising, and premised on the belief that the FDA lacks such jurisdiction under the FDCA. As a result, Congress' tobacco-specific statutes preclude the FDA from regulating tobacco products as customarily marketed.

Although the dissent takes issue with our discussion of the FDA's change in position, post, at 186-189, our conclusion does not rely on the fact that the FDA's assertion of jurisdiction represents a sharp break with its prior interpretation of the FDCA. Certainly, an agency's initial interpretation of a statute that it is charged with administering is not "carved [157] in stone." Chevron, 467 U. S., at 863; see also Smiley v. Citibank (South Dakota), N. A., 517 U. S. 735, 742 (1996). As we recognized in Motor Vehicle Mfrs. Assn. of United States, Inc. v. State Farm Mut. Automobile Ins. Co., 463 U. S. 29 (1983), agencies "must be given ample latitude to `adapt their rules and policies to the demands of changing circumstances.' " Id., at 42 (quoting Permian Basin Area Rate Cases, 390 U. S. 747, 784 (1968)). The consistency of the FDA's prior position is significant in this case for a different reason: It provides important context to Congress' enactment of its tobacco-specific legislation. When the FDA repeatedly informed Congress that the FDCA does not grant it the authority to regulate tobacco products, its statements were consistent with the agency's unwavering position since its inception, and with the position that its predecessor agency had first taken in 1914. Although not crucial, the consistency of the FDA's prior position bolsters the conclusion that when Congress created a distinct regulatory scheme addressing the subject of tobacco and health, it understood that the FDA is without jurisdiction to regulate tobacco products and ratified that position.

The dissent also argues that the proper inference to be drawn from Congress' tobacco-specific legislation is "critically ambivalent." Post, at 182. We disagree. In that series of statutes, Congress crafted a specific legislative response to the problem of tobacco and health, and it did so with the understanding, based on repeated assertions by the FDA, that the agency has no authority under the FDCA to regulate tobacco products. Moreover, Congress expressly pre-empted any other regulation of the labeling of tobacco products concerning their health consequences, even though the oversight of labeling is central to the FDCA's regulatory scheme. And in addressing the subject, Congress consistently evidenced its intent to preclude any federal agency from exercising significant policymaking authority in the area. Under these circumstances, we believe the appropriate [158] inference—that Congress intended to ratify the FDA's prior position that it lacks jurisdiction—is unmistakable.

The dissent alternatively argues that, even if Congress' subsequent tobacco-specific legislation did, in fact, ratify the FDA's position, that position was merely a contingent disavowal of jurisdiction. Specifically, the dissent contends that "the FDA's traditional view was largely premised on a perceived inability to prove the necessary statutory `intent' requirement." Post, at 189-190. A fair reading of the FDA's representations prior to 1995, however, demonstrates that the agency's position was essentially unconditional. See, e. g., 1972 Hearings 239, 242 (statement of Comm'r Edwards) ("[R]egulation of cigarettes is to be the domain of Congress," and "[a]ny such move by FDA would be inconsistent with the clear congressional intent"); 1983 House Hearings 74 (statement of Assistant Secretary Brandt) ("[T]he issue of regulation of tobacco . . . is something that Congress has reserved to itself"); 1983 Senate Hearings 56 (statement of Assistant Secretary Brandt) ("Congress has assumed the responsibility of regulating . . . cigarettes"); Brief for Appellee in Action on Smoking and Health v. Harris, 655 F. 2d 236 (CADC 1980), 9 Rec. in No. 97-1604 (CA4), Tab No. 4, at 27, n. 23 (because "Congress has never acted to disturb the agency's interpretation," it "acquiesced in the FDA's interpretation"). To the extent the agency's position could be characterized as equivocal, it was only with respect to the well-established exception of when the manufacturer makes express claims of therapeutic benefit. See, e. g., 1965 Hearings 193 (statement of Deputy Comm'r Rankin) ("The Food and Drug Administration has no jurisdiction under the Food, Drug, and Cosmetic Act over tobacco, unless it bears drug claims"); Letter to ASH Executive Director Banzhaf from FDA Comm'r Kennedy (Dec. 5, 1977), App. 47 ("The interpretation of the Act by FDA consistently has been that cigarettes are not a drug unless health claims are made by the vendors"); Letter to ASH Executive Director Banzhaf from [159] FDA Comm'r Goyan (Nov. 25, 1980), id., at 67 ("Insofar as rulemaking would relate to cigarettes or attached filters as customarily marketed, we have concluded that FDA has no jurisdiction"). Thus, what Congress ratified was the FDA's plain and resolute position that the FDCA gives the agency no authority to regulate tobacco products as customarily marketed.

Finally, our inquiry into whether Congress has directly spoken to the precise question at issue is shaped, at least in some measure, by the nature of the question presented. Deference under Chevron to an agency's construction of a statute that it administers is premised on the theory that a statute's ambiguity constitutes an implicit delegation from Congress to the agency to fill in the statutory gaps. See Chevron, supra, at 844. In extraordinary cases, however, there may be reason to hesitate before concluding that Congress has intended such an implicit delegation. Cf. Breyer, Judicial Review of Questions of Law and Policy, 38 Admin. L. Rev. 363, 370 (1986) ("A court may also ask whether the legal question is an important one. Congress is more likely to have focused upon, and answered, major questions, while leaving interstitial matters to answer themselves in the course of the statute's daily administration").

This is hardly an ordinary case. Contrary to its representations to Congress since 1914, the FDA has now asserted jurisdiction to regulate an industry constituting a significant portion of the American economy. In fact, the FDA contends that, were it to determine that tobacco products provide no "reasonable assurance of safety," it would have the authority to ban cigarettes and smokeless tobacco entirely. See Brief for Petitioners 35-36; Reply Brief for Petitioners 14. Owing to its unique place in American history and society, tobacco has its own unique political history. Congress, for better or for worse, has created a distinct regulatory scheme for tobacco products, squarely rejected proposals to [160] give the FDA jurisdiction over tobacco, and repeatedly acted to preclude any agency from exercising significant policymaking authority in the area. Given this history and the breadth of the authority that the FDA has asserted, we are obliged to defer not to the agency's expansive construction of the statute, but to Congress' consistent judgment to deny the FDA this power.

Our decision in MCI Telecommunications Corp. v. American Telephone & Telegraph Co., 512 U. S. 218 (1994), is instructive. That case involved the proper construction of the term "modify" in § 203(b) of the Communications Act of 1934. The FCC contended that, because the Act gave it the discretion to "modify any requirement" imposed under the statute, it therefore possessed the authority to render voluntary the otherwise mandatory requirement that long distance carriers file their rates. Id., at 225. We rejected the FCC's construction, finding "not the slightest doubt" that Congress had directly spoken to the question. Id., at 228. In reasoning even more apt here, we concluded that "[i]t is highly unlikely that Congress would leave the determination of whether an industry will be entirely, or even substantially, rateregulated to agency discretion—and even more unlikely that it would achieve that through such a subtle device as permission to `modify' rate-filing requirements." Id., at 231.

As in MCI, we are confident that Congress could not have intended to delegate a decision of such economic and political significance to an agency in so cryptic a fashion. To find that the FDA has the authority to regulate tobacco products, one must not only adopt an extremely strained understanding of "safety" as it is used throughout the Act—a concept central to the FDCA's regulatory scheme—but also ignore the plain implication of Congress' subsequent tobaccospecific legislation. It is therefore clear, based on the FDCA's overall regulatory scheme and the subsequent tobacco legislation, that Congress has directly spoken to the [161] question at issue and precluded the FDA from regulating tobacco products.

* * *

By no means do we question the seriousness of the problem that the FDA has sought to address. The agency has amply demonstrated that tobacco use, particularly among children and adolescents, poses perhaps the single most significant threat to public health in the United States. Nonetheless, no matter how "important, conspicuous, and controversial" the issue, and regardless of how likely the public is to hold the Executive Branch politically accountable, post, at 190, an administrative agency's power to regulate in the public interest must always be grounded in a valid grant of authority from Congress. And "`[i]n our anxiety to effectuate the congressional purpose of protecting the public, we must take care not to extend the scope of the statute beyond the point where Congress indicated it would stop.' " United States v. Article of Drug . . . Bacto-Unidisk, 394 U. S. 784, 800 (1969) (quoting 62 Cases of Jam v. United States, 340 U. S. 593, 600 (1951)). Reading the FDCA as a whole, as well as in conjunction with Congress' subsequent tobaccospecific legislation, it is plain that Congress has not given the FDA the authority that it seeks to exercise here. For these reasons, the judgment of the Court of Appeals for the Fourth Circuit is affirmed.

It is so ordered.

The Food and Drug Administration (FDA) has the authority to regulate "articles (other than food) intended to affect the structure or any function of the body . . . ." Federal Food, Drug, and Cosmetic Act (FDCA), 21 U. S. C. § 321(g)(1)(C). Unlike the majority, I believe that tobacco products fit within this statutory language.

[162] In its own interpretation, the majority nowhere denies the following two salient points. First, tobacco products (including cigarettes) fall within the scope of this statutory definition, read literally. Cigarettes achieve their moodstabilizing effects through the interaction of the chemical nicotine and the cells of the central nervous system. Both cigarette manufacturers and smokers alike know of, and desire, that chemically induced result. Hence, cigarettes are "intended to affect" the body's "structure" and "function," in the literal sense of these words.

Second, the statute's basic purpose—the protection of public health—supports the inclusion of cigarettes within its scope. See United States v. Article of Drug . . . BactoUnidisk, 394 U. S. 784, 798 (1969) (FDCA "is to be given a liberal construction consistent with [its] overriding purpose to protect the public health " (emphasis added)). Unregulated tobacco use causes "[m]ore than 400,000 people [to] die each year from tobacco-related illnesses, such as cancer, respiratory illnesses, and heart disease." 61 Fed. Reg. 44398 (1996). Indeed, tobacco products kill more people in this country every year "than . . . AIDS . . . , car accidents, alcohol, homicides, illegal drugs, suicides, and fires, combined. " Ibid. (emphasis added).

Despite the FDCA's literal language and general purpose (both of which support the FDA's finding that cigarettes come within its statutory authority), the majority nonetheless reads the statute as excluding tobacco products for two basic reasons:

(1) the FDCA does not "fit" the case of tobacco because the statute requires the FDA to prohibit dangerous drugs or devices (like cigarettes) outright, and the agency concedes that simply banning the sale of cigarettes is not a proper remedy, ante, at 139-141; and (2) Congress has enacted other statutes, which, when viewed in light of the FDA's long history of denying [163] tobacco-related jurisdiction and considered together with Congress' failure explicitly to grant the agency tobacco-specific authority, demonstrate that Congress did not intend for the FDA to exercise jurisdiction over tobacco, ante, at 155-156.

In my view, neither of these propositions is valid. Rather, the FDCA does not significantly limit the FDA's remedial alternatives. See infra, at 174-181. And the later statutes do not tell the FDA it cannot exercise jurisdiction, but simply leave FDA jurisdictional law where Congress found it. See infra, at 181-186; cf. Food and Drug Administration Modernization Act of 1997, 111 Stat. 2380 (codified at note following 21 U. S. C. § 321 (1994 ed., Supp. III)) (statute "shall" not "be construed to affect the question of whether" the FDA "has any authority to regulate any tobacco product").

The bulk of the opinion that follows will explain the basis for these latter conclusions. In short, I believe that the most important indicia of statutory meaning—language and purpose—along with the FDCA's legislative history (described briefly in Part I) are sufficient to establish that the FDA has authority to regulate tobacco. The statute-specific arguments against jurisdiction that the tobacco companies and the majority rely upon (discussed in Part II) are based on erroneous assumptions and, thus, do not defeat the jurisdiction-supporting thrust of the FDCA's language and purpose. The inferences that the majority draws from later legislative history are not persuasive, since (as I point out in Part III) one can just as easily infer from the later laws that Congress did not intend to affect the FDA's tobacco-related authority at all. And the fact that the FDA changed its mind about the scope of its own jurisdiction is legally insignificant because (as Part IV establishes) the agency's reasons for changing course are fully justified. Finally, as I explain in Part V, the degree of accountability that likely will attach to the FDA's action in this case should alleviate any concern [164] that Congress, rather than an administrative agency, ought to make this important regulatory decision.

Before 1938, the federal Pure Food and Drug Act contained only two jurisdictional definitions of "drug":

"[1] medicines and preparations recognized in the United States Pharmacopoeia or National Formulary . . . and [2] any substance or mixture of substances intended to be used for the cure, mitigation, or prevention of disease." Act of June 30, 1906, ch. 3915, § 6, 34 Stat. 769.

In 1938, Congress added a third definition, relevant here:

"(3) articles (other than food) intended to affect the structure or any function of the body . . . ." Act of June 25, 1938, ch. 675, § 201(g), 52 Stat. 1041 (codified at 21 U. S. C. § 321(g)(1)(C)).

It also added a similar definition in respect to a "device." See § 201(h), 52 Stat. 1041 (codified at 21 U. S. C. § 321(h)). As I have mentioned, the literal language of the third definition and the FDCA's general purpose both strongly support a projurisdiction reading of the statute. See supra, at 161-162.

The statute's history offers further support. The FDA drafted the new language, and it testified before Congress that the third definition would expand the FDCA's jurisdictional scope significantly. See Hearings on S. 1944 before a Subcommittee of the Senate Committee on Commerce, 73d Cong., 2d Sess., 15-16 (1933), reprinted in 1 FDA, Legislative History of the Federal Food, Drug, and Cosmetic Act and Its Amendments 107-108 (1979) (hereinafter Leg. Hist.). Indeed, "[t]he purpose" of the new definition was to "make possible the regulation of a great many products that have been found on the market that cannot be alleged to be treatments for diseased conditions." Id., at 108. While the drafters focused specifically upon the need to give the FDA jurisdiction [165] over "slenderizing" products such as "antifat remedies," ibid., they were aware that, in doing so, they had created what was "admittedly an inclusive, a wide definition," id., at 107. And that broad language was included deliberately, so that jurisdiction could be had over "all substances and preparations, other than food, and all devices intended to affect the structure or any function of the body . . . ." Ibid. (emphasis added); see also Hearings on S. 2800 before the Senate Committee on Commerce, 73d Cong., 2d Sess., 516 (1934), reprinted in 2 Leg. Hist. 519 (statement of then-FDA Chief Walter Campbell acknowledging that "[t]his definition of `drugs' is all-inclusive").

After studying the FDCA's history, experts have written that the statute "is a purposefully broad delegation of discretionary powers by Congress," 1 J. O'Reilly, Food and Drug Administration § 6.01, p. 6-1 (2d ed. 1995) (hereinafter O'Reilly), and that, in a sense, the FDCA "must be regarded as a constitution " that "establish[es] general principles" and "permit[s] implementation within broad parameters" so that the FDA can "implement these objectives through the most effective and efficient controls that can be devised." Hutt, Philosophy of Regulation Under the Federal Food, Drug and Cosmetic Act, 28 Food Drug Cosm. L. J. 177, 178-179 (1973) (emphasis added). This Court, too, has said that the

"historical expansion of the definition of drug, and the creation of a parallel concept of devices, clearly show . . . that Congress fully intended that the Act's coverage be as broad as its literal language indicates—and equally clearly, broader than any strict medical definition might otherwise allow." Bacto-Unidisk, 394 U. S., at 798.

That Congress would grant the FDA such broad jurisdictional authority should surprise no one. In 1938, the President and much of Congress believed that federal administrative agencies needed broad authority and would exercise that authority wisely—a view embodied in much Second New [166] Deal legislation. Cf. Gray v. Powell, 314 U. S. 402, 411-412 (1941) (Congress "could have legislated specifically" but decided "to delegate that function to those whose experience in a particular field gave promise of a better informed, more equitable" determination). Thus, at around the same time that it added the relevant language to the FDCA, Congress enacted laws granting other administrative agencies even broader powers to regulate much of the Nation's transportation and communication. See, e. g., Civil Aeronautics Act of 1938, ch. 601, § 401(d)(1), 52 Stat. 987 (Civil Aeronautics Board to regulate airlines within confines of highly general "public convenience and necessity" standard); Motor Carrier Act of 1935, ch. 498, § 204(a)(1), 49 Stat. 546 (Interstate Commerce Commission to establish "reasonable requirements" for trucking); Communications Act of 1934, ch. 652, § 201(a), 48 Stat. 1070 (Federal Communications Commission (FCC) to regulate radio, later television, within confines of even broader "public interest" standard). Why would the 1938 New Deal Congress suddenly have hesitated to delegate to so well established an agency as the FDA all of the discretionary authority that a straightforward reading of the relevant statutory language implies?

Nor is it surprising that such a statutory delegation of power could lead after many years to an assertion of jurisdiction that the 1938 legislators might not have expected. Such a possibility is inherent in the very nature of a broad delegation. In 1938, it may well have seemed unlikely that the FDA would ever bring cigarette manufacturers within the FDCA's statutory language by proving that cigarettes produce chemical changes in the body and that the makers "intended" their product chemically to affect the body's "structure" or "function." Or, back then, it may have seemed unlikely that, even assuming such proof, the FDA actually would exercise its discretion to regulate so popular a product. See R. Kluger, Ashes to Ashes 105 (1997) (in the 1930's "Americans were in love with smoking . . .").

[167] But it should not have seemed unlikely that, assuming the FDA decided to regulate and proved the particular jurisdictional prerequisites, the courts would rule such a jurisdictional assertion fully authorized. Cf. United States v. Southwestern Cable Co., 392 U. S. 157, 172 (1968) (reading Communications Act of 1934 as authorizing FCC jurisdiction to regulate cable systems while noting that "Congress could not in 1934 have foreseen the development of" advanced communications systems). After all, this Court has read more narrowly phrased statutes to grant what might have seemed even more unlikely assertions of agency jurisdiction. See, e. g., Permian Basin Area Rate Cases, 390 U. S. 747, 774-777 (1968) (statutory authority to regulate interstate "transportation" of natural gas includes authority to regulate "prices" charged by field producers); Phillips Petroleum Co. v. Wisconsin, 347 U. S. 672, 677-684 (1954) (independent gas producer subject to regulation despite Natural Gas Act's express exemption of gathering and production facilities).

I shall not pursue these general matters further, for neither the companies nor the majority denies that the FDCA's literal language, its general purpose, and its particular legislative history favor the FDA's present jurisdictional view. Rather, they have made several specific arguments in support of one basic contention: Even if the statutory delegation is broad, it is not broad enough to include tobacco. I now turn to each of those arguments.

The tobacco companies contend that the FDCA's words cannot possibly be read to mean what they literally say. The statute defines "device," for example, as "an instrument, apparatus, implement, machine, contrivance, implant, in vitro reagent, or other similar or related article . . . intended to affect the structure or any function of the body . . . ." 21 [168] U. S. C. § 321(h). Taken literally, this definition might include everything from room air conditioners to thermal pajamas. The companies argue that, to avoid such a result, the meaning of "drug" or "device" should be confined to medical or therapeutic products, narrowly defined. See Brief for Respondent United States Tobacco Co. 8-9.

The companies may well be right that the statute should not be read to cover room air conditioners and winter underwear. But I do not agree that we must accept their proposed limitation. For one thing, such a cramped reading contravenes the established purpose of the statutory language. See Bacto-Unidisk, 394 U. S., at 798 (third definition is "clearly, broader than any strict medical definition"); 1 Leg. Hist. 108 (definition covers products "that cannot be alleged to be treatments for diseased conditions"). For another, the companies' restriction would render the other two "drug" definitions superfluous. See 21 U. S. C. §§ 321(g)(1)(A), (g)(1)(B) (covering articles in the leading pharmacology compendia and those "intended for use in the diagnosis, cure, mitigation, treatment, or prevention of disease").

Most importantly, the statute's language itself supplies a different, more suitable, limitation: that a "drug" must be a chemical agent. The FDCA's "device" definition states that an article which affects the structure or function of the body is a "device" only if it "does not achieve its primary intended purposes through chemical action within . . . the body," and "is not dependent upon being metabolized for the achievement of its primary intended purposes." § 321(h) (emphasis added). One can readily infer from this language that at least an article that does achieve its primary purpose through chemical action within the body and that is dependent upon being metabolized is a "drug," provided that it otherwise falls within the scope of the "drug" definition. And one need not hypothesize about air conditioners or thermal [169] pajamas to recognize that the chemical nicotine, an important tobacco ingredient, meets this test.

Although I now oversimplify, the FDA has determined that once nicotine enters the body, the blood carries it almost immediately to the brain. See 61 Fed. Reg. 44698-44699 (1966). Nicotine then binds to receptors on the surface of brain cells, setting off a series of chemical reactions that alter one's mood and produce feelings of sedation and stimulation. See id., at 44699, 44739. Nicotine also increases the number of nicotinic receptors on the brain's surface, and alters its normal electrical activity. See id., at 44739. And nicotine stimulates the transmission of a natural chemical that "rewards" the body with pleasurable sensations (dopamine), causing nicotine addiction. See id., at 44700, 44721-44722. The upshot is that nicotine stabilizes mood, suppresses appetite, tranquilizes, and satisfies a physical craving that nicotine itself has helped to create—all through chemical action within the body after being metabolized.

This physiology—and not simply smoker psychology— helps to explain why as many as 75% of adult smokers believe that smoking "reduce[s] nervous irritation," 60 Fed. Reg. 41579 (1995); why 73% of young people (10- to 22-yearolds) who begin smoking say they do so for "relaxation," 61 Fed. Reg. 44814 (1996); and why less than 3% of smokers succeed in quitting each year, although 70% want to quit, id., at 44704. That chemistry also helps to explain the Surgeon General's findings that smokers believe "smoking [makes them] feel better" and smoke more "in situations involving negative mood." Id., at 44814. And, for present purposes, that chemistry demonstrates that nicotine affects the "structure" and "function" of the body in a manner that is quite similar to the effects of other regulated substances. See id., at 44667 (FDA regulates Valium, NoDoz, weight-loss products). Indeed, addiction, sedation, stimulation, and weight loss are precisely the kinds of product effects that the FDA typically reviews and controls. And, since the nicotine in cigarettes [170] plainly is not a "food," its chemical effects suffice to establish that it is as a "drug" (and the cigarette that delivers it a drug-delivery "device") for the purpose of the FDCA.

The tobacco companies' principal definitional argument focuses upon the statutory word "intended." See 21 U. S. C. § 321(g)(1)(C). The companies say that "intended" in this context is a term of art. See Brief for Respondent Brown & Williamson Tobacco Corp. 2. They assert that the statutory word "intended" means that the product's maker has made an express claim about the effect that its product will have on the body. Ibid. Indeed, according to the companies, the FDA's inability to prove that cigarette manufacturers make such claims is precisely why that agency historically has said it lacked the statutory power to regulate tobacco. See id., at 19-20.

The FDCA, however, does not use the word "claimed"; it uses the word "intended." And the FDA long ago issued regulations that say the relevant "intent" can be shown not only by a manufacturer's "expressions," but also "by the circumstances surrounding the distribution of the article." 41 Fed. Reg. 6896 (1976) (codified at 21 CFR § 801.4 (1999)); see also 41 Fed. Reg. 6896 (1976) ("objective intent" shown if "article is, with the knowledge [of its makers], offered and used" for a particular purpose). Thus, even in the absence of express claims, the FDA has regulated products that affect the body if the manufacturer wants, and knows, that consumers so use the product. See, e. g., 60 Fed. Reg. 41527-41531 (1995) (describing agency's regulation of topical hormones, sunscreens, fluoride, tanning lamps, thyroid in food supplements, novelty condoms—all marketed without express claims); see also 1 O'Reilly § 13.04, at 13-15 ("Sometimes the very nature of the material makes it a drug . . .").

Courts ordinarily reverse an agency interpretation of this kind only if Congress has clearly answered the interpretive [171] question or if the agency's interpretation is unreasonable. Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837, 842-843 (1984). The companies, in an effort to argue the former, point to language in the legislative history tying the word "intended" to a technical concept called "intended use." But nothing in Congress' discussion either of "intended" or "intended use" suggests that an express claim (which often shows intent) is always necessary. Indeed, the primary statement to which the companies direct our attention says only that a manufacturer can determine what kind of regulation applies—"food" or "drug"—because, "through his representations in connection with its sale, [the manufacturer] can determine" whether an article is to be used as a "food," as a "drug," or as "both." S. Rep. No. 361, 74th Cong., 1st Sess., 4 (1935), reprinted in 3 Leg. Hist. 696.

Nor is the FDA's "objective intent" interpretation unreasonable. It falls well within the established scope of the ordinary meaning of the word "intended." See Agnew v. United States, 165 U. S. 36, 53 (1897) (intent encompasses the known consequences of an act). And the companies acknowledge that the FDA can regulate a drug-like substance in the ordinary circumstance, i. e., where the manufacturer makes an express claim, so it is not unreasonable to conclude that the agency retains such power where a product's effects on the body are so well known (say, like those of aspirin or calamine lotion), that there is no need for express representations because the product speaks for itself.

The companies also cannot deny that the evidence of their intent is sufficient to satisfy the statutory word "intended" as the FDA long has interpreted it. In the first place, there was once a time when they actually did make express advertising claims regarding tobacco's mood-stabilizing and weight-reducing properties—and historical representations can portend present expectations. In the late 1920's, for example, the American Tobacco Company urged weightconscious smokers to "`Reach for a Lucky instead of a [172] sweet.' " Kluger, Ashes to Ashes, at 77-78. The advertisements of R J Reynolds (RJR) emphasized mood stability by depicting a pilot remarking that "`It Takes Steady Nerves to Fly the Mail at Night . . . . That's why I smoke Camels. And I smoke plenty!' " Id., at 86. RJR also advertised the stimulating quality of cigarettes, stating in one instance that "`You get a Lift with a Camel,' " and, in another, that Camels are "`A Harmless Restoration of the Flow of Natural Body Energy.' " Id., at 87. And claims of medical proof of mildness (and of other beneficial effects) once were commonplace. See, e. g., id., at 93 (Brown & Williamson advertised Koolbrand mentholated cigarettes as "a tonic to hot, tired throats"); id., at 101, 131 (Philip Morris contended that "`[r]ecognized laboratory tests have conclusively proven the advantage of Phillip [sic] Morris' "); id., at 88 (RJR proclaimed "`For Digestion's sake, smoke Camels! . . . Camels make mealtime more pleasant—digestion is stimulated—alkalinity increased' "). Although in recent decades cigarette manufacturers have stopped making express health claims in their advertising, consumers have come to understand what the companies no longer need to express—that through chemical action cigarettes stabilize mood, sedate, stimulate, and help suppress appetite.

Second, even though the companies refused to acknowledge publicly (until only very recently) that the nicotine in cigarettes has chemically induced, and habit-forming, effects, see, e. g., Regulation of Tobacco Products (Part 1): Hearings before the House Subcommittee on Health and the Environment, 103d Cong., 2d Sess., 628 (1994) (hereinafter 1994 Hearings) (heads of seven major tobacco companies testified under oath that they believed "nicotine is not addictive" (emphasis added)), the FDA recently has gained access to solid, documentary evidence proving that cigarette manufacturers have long known tobacco produces these effects within the body through the metabolizing of chemicals, and that they [173] have long wanted their products to produce those effects in this way.

For example, in 1972, a tobacco-industry scientist explained that "`[s]moke is beyond question the most optimized vehicle of nicotine,' " and "`the cigarette is the most optimized dispenser of smoke.' " 61 Fed. Reg. 44856 (1996) (emphasis deleted). That same scientist urged company executives to

"`[t]hink of the cigarette pack as a storage container for a day's supply of nicotine. . . . Think of the cigarette as a dispenser for a dose unit of nicotine [and] [t]hink of a puff of smoke as the vehicle of nicotine.' " Ibid. (Philip Morris) (emphasis deleted).

That same year, other tobacco industry researchers told their superiors that

"`in different situations and at different dose levels, nicotine appears to act as a stimulant, depressant, tranquilizer, psychic energizer, appetite reducer, anti-fatigue agent, or energizer. . . . Therefore, [tobacco] products may, in a sense, compete with a variety of other products with certain types of drug action.' " Id., at 44669 (RJR) (emphasis deleted).

A draft report prepared by authorities at Philip Morris said that nicotine

"`is a physiologically active, nitrogen containing substance [similar to] quinine, cocaine, atropine and morphine. [And] [w]hile each of these [other] substances can be used to affect human physiology, nicotine has a particularly broad range of influence.' " Id., at 44668-44669.

And a 1980 manufacturer's study stated that

"`the pharmacological response of smokers to nicotine is believed to be responsible for an individual's smoking [174] behaviour, providing the motivation for and the degree of satisfaction required by the smoker.' " Id., at 44936 (Brown & Williamson).

With such evidence, the FDA has more than sufficiently established that the companies "intend" their products to "affect" the body within the meaning of the FDCA.

The majority nonetheless reaches the "inescapable conclusion" that the language and structure of the FDCA as a whole "simply do not fit" the kind of public health problem that tobacco creates. Ante, at 143. That is because, in the majority's view, the FDCA requires the FDA to ban outright "dangerous" drugs or devices (such as cigarettes); yet, the FDA concedes that an immediate and total cigarette-sale ban is inappropriate. Ibid.

This argument is curious because it leads with similarly "inescapable" force to precisely the opposite conclusion, namely, that the FDA does have jurisdiction but that it must ban cigarettes. More importantly, the argument fails to take into account the fact that a statute interpreted as requiring the FDA to pick a more dangerous over a less dangerous remedy would be a perverse statute, causing, rather than preventing, unnecessary harm whenever a total ban is likely the more dangerous response. And one can at least imagine such circumstances.

Suppose, for example, that a commonly used, mildly addictive sleeping pill (or, say, a kind of popular contact lens), plainly within the FDA's jurisdiction, turned out to pose serious health risks for certain consumers. Suppose further that many of those addicted consumers would ignore an immediate total ban, turning to a potentially more dangerous black-market substitute, while a less draconian remedy (say, adequate notice) would wean them gradually away to a safer product. Would the FDCA still force the FDA to impose [175] the more dangerous remedy? For the following reasons, I think not.

First, the statute's language does not restrict the FDA's remedial powers in this way. The FDCA permits the FDA to regulate a "combination product"—i. e., a "device" (such as a cigarette) that contains a "drug" (such as nicotine)— under its "device" provisions. 21 U. S. C. § 353(g)(1). And the FDCA's "device" provisions explicitly grant the FDA wide remedial discretion. For example, where the FDA cannot "otherwise" obtain "reasonable assurance" of a device's "safety and effectiveness," the agency may restrict by regulation a product's "sale, distribution, or use" upon "such . . . conditions as the Secretary may prescribe." § 360j(e)(1) (emphasis added). And the statutory section that most clearly addresses the FDA's power to ban (entitled "Banned devices") says that, where a device presents "an unreasonable and substantial risk of illness or injury," the Secretary "may "—not must —"initiate a proceeding . . . to make such device a banned device." § 360f(a) (emphasis added).

The Court points to other statutory subsections which it believes require the FDA to ban a drug or device entirely, even where an outright ban risks more harm than other regulatory responses. See ante, at 135-136. But the cited provisions do no such thing. It is true, as the majority contends, that "the FDCA requires the FDA to place all devices" in "one of three classifications" and that Class III devices require "premarket approval." Ante, at 136. But it is not the case that the FDA must place cigarettes in Class III because tobacco itself "presents a potential unreasonable risk of illness or injury." 21 U. S. C. § 360c(a)(1)(C). In fact, Class III applies only where regulation cannot otherwise "provide reasonable assurance of . . . safety." §§ 360c(a) (1)(A), (B) (placing a device in Class I or Class II when regulation can provide that assurance). Thus, the statute plainly allows the FDA to consider the relative, overall "safety" of [176] a device in light of its regulatory alternatives, and where the FDA has chosen the least dangerous path, i. e., the safest path, then it can—and does—provide a "reasonable assurance" of "safety" within the meaning of the statute. A good football helmet provides a reasonable assurance of safety for the player even if the sport itself is still dangerous. And the safest regulatory choice by definition offers a "reasonable" assurance of safety in a world where the other alternatives are yet more dangerous.

In any event, it is not entirely clear from the statute's text that a Class III categorization would require the FDA affirmatively to withdraw from the market dangerous devices, such as cigarettes, which are already widely distributed. See, e. g., § 360f(a) (when a device presents an "unreasonable and substantial risk of illness or injury," the Secretary "may" make it "a banned device"); § 360h(a) (when a device "presents an unreasonable risk of substantial harm to the public health," the Secretary "may" require "notification"); § 360h(b) (when a defective device creates an "unreasonable risk" of harm, the Secretary "may" order "[r]epair, replacement, or refund"); cf. 2 O'Reilly § 18.08, at 18-29 (point of Class III "premarket approval" is to allow "careful scientific review" of each "truly new" device "before it is exposed" to users (emphasis added)).

Noting that the FDCA requires banning a "misbranded" drug, the majority also points to 21 U. S. C. § 352(j), which deems a drug or device "misbranded" if "it is dangerous to health when used" as "prescribed, recommended, or suggested in the labeling." See ante, at 135. In addition, the majority mentions § 352(f)(1), which calls a drug or device "misbranded" unless "its labeling bears . . . adequate directions for use" as "are necessary for the protection of users." Ibid. But this "misbranding" language is not determinative, for it permits the FDA to conclude that a drug or device is not "dangerous to health" and that it does have "adequate" [177] directions when regulated so as to render it as harmless as possible. And surely the agency can determine that a substance is comparatively "safe" (not "dangerous") whenever it would be less dangerous to make the product available (subject to regulatory requirements) than suddenly to withdraw it from the market. Any other interpretation risks substantial harm of the sort that my sleeping pill example illustrates. See supra, at 174-175. And nothing in the statute prevents the agency from adopting a view of "safety" that would avoid such harm. Indeed, the FDA already seems to have taken this position when permitting distribution of toxic drugs, such as poisons used for chemotherapy, that are dangerous for the user but are not deemed "dangerous to health" in the relevant sense. See 61 Fed. Reg. 44413 (1996).

The tobacco companies point to another statutory provision which says that if a device "would cause serious, adverse health consequences or death, the Secretary shall issue" a cease distribution order. 21 U. S. C. § 360h(e)(1) (emphasis added). But that word "shall" in this context cannot mean that the Secretary must resort to the recall remedy whenever a device would have serious, adverse health effects. Rather, that language must mean that the Secretary "shall issue" a cease distribution order in compliance with the section's procedural requirements if the Secretary chooses in her discretion to use that particular subsection's recall remedy. Otherwise, the subsection would trump and make meaningless the same section's provision of other lesser remedies such as simple "notice" (which the Secretary similarly can impose if, but only if, she finds that the device "presents an unreasonable risk of substantial harm to the public"). § 360h(a)(1). And reading the statute to compel the FDA to "recall" every dangerous device likewise would conflict with that same subsection's statement that the recall remedy "shall be in addition to [the other] remedies provided" in the statute. § 360h(e)(3) (emphasis added).

[178] The statute's language, then, permits the agency to choose remedies consistent with its basic purpose—the overall protection of public health.

The second reason the FDCA does not require the FDA to select the more dangerous remedy, see supra, at 175, is that, despite the majority's assertions to the contrary, the statute does not distinguish among the kinds of health effects that the agency may take into account when assessing safety. The Court insists that the statute only permits the agency to take into account the health risks and benefits of the "product itself " as used by individual consumers, ante, at 140, and, thus, that the FDA is prohibited from considering that a ban on smoking would lead many smokers to suffer severe withdrawal symptoms or to buy possibly stronger, more dangerous, black market cigarettes—considerations that the majority calls "the aggregate health effects of alternative administrative actions." Ibid. But the FDCA expressly permits the FDA to take account of comparative safety in precisely this manner. See, e. g., 21 U. S. C. § 360h(e)(2)(B)(i)(II) (no device recall if "risk of recal[l]" presents "a greater health risk than" no recall); § 360h(a) (notification "unless" notification "would present a greater danger" than "no such notification").

Moreover, one cannot distinguish in this context between a "specific" health risk incurred by an individual and an "aggregate" risk to a group. All relevant risk is, at bottom, risk to an individual; all relevant risk attaches to "the product itself"; and all relevant risk is "aggregate" in the sense that the agency aggregates health effects in order to determine risk to the individual consumer. If unregulated smoking will kill 4 individuals out of a typical group of 1,000 people, if regulated smoking will kill 1 out of 1,000, and if a smoking ban (because of the black market) will kill 2 out of 1,000; then these three possibilities mean that in each group four, one, and two individuals, on average, will die respectively. And the risk to each individual consumer is 4/1,000, [179] 1/1,000, and 2/1,000 respectively. A "specific" risk to an individual consumer and "aggregate" risks are two sides of the same coin; each calls attention to the same set of facts. While there may be a theoretical distinction between the risk of the product itself and the risk related to the presence or absence of an intervening voluntary act (e. g., the search for a replacement on the black market), the majority does not rely upon any such distinction, and the FDA's history of regulating "replacement" drugs such as methadone shows that it has long taken likely actual alternative consumer behavior into account.

I concede that, as a matter of logic, one could consider the FDA's "safety" evaluation to be different from its choice of remedies. But to read the statute to forbid the agency from taking account of the realities of consumer behavior either in assessing safety or in choosing a remedy could increase the risks of harm—doubling the risk of death to each "individual user" in my example above. Why would Congress insist that the FDA ignore such realities, even if the consequent harm would occur only unusually, say, where the FDA evaluates a product (a sleeping pill; a cigarette; a contact lens) that is already on the market, potentially habit forming, or popular? I can find no satisfactory answer to this question. And that, I imagine, is why the statute itself says nothing about any of the distinctions that the Court has tried to draw. See 21 U. S. C. § 360c(a)(2) (instructing FDA to determine the safety and effectiveness of a "device" in part by weighing "any probable benefit to health . . . against any probable risk of injury or illness . . ." (emphasis added)).

Third, experience counsels against an overly rigid interpretation of the FDCA that is divorced from the statute's overall health-protecting purposes. A different set of words, added to the FDCA in 1958 by the Delaney Amendment, provides that "no [food] additive shall be deemed to be safe if it is found [after appropriate tests] to induce cancer when ingested by man or animal." § 348(c)(3). The FDA [180] once interpreted this language as requiring it to ban any food additive, no matter how small the amount, that appeared in any food product if that additive was ever found to induce cancer in any animal, no matter how large a dose needed to induce the appearance of a single carcinogenic cell. See H. R. Rep. No. 95-658, p. 7 (1977) (discussing agency's view). The FDA believed that the statute's ban mandate was absolute and prevented it from establishing a level of "safe use" or even to judge whether "the benefits of continued use outweigh the risks involved." Id., at 5. This interpretation— which in principle could have required the ban of everything from herbal teas to mushrooms—actually led the FDA to ban saccharine, see 42 Fed. Reg. 19996 (1977), though this extremely controversial regulatory response never took effect because Congress enacted, and has continually renewed, a law postponing the ban. See Saccharin Study and Labeling Act, Pub. L. 95-203, § 3, 91 Stat. 1452; e. g., Pub. L. 102-142, Tit. VI, 105 Stat. 910.

The Court's interpretation of the statutory language before us risks Delaney-type consequences with even less linguistic reason. Even worse, the view the Court advances undermines the FDCA's overall health-protecting purpose by placing the FDA in the strange dilemma of either banning completely a potentially dangerous drug or device or doing nothing at all. Saying that I have misunderstood its conclusion, the majority maintains that the FDA "may clearly regulate many `dangerous' products without banning them." Ante, at 142. But it then adds that the FDA must ban— rather than otherwise regulate—a drug or device that "cannot be used safely for any therapeutic purpose." Ibid. If I misunderstand, it is only because this linchpin of the majority's conclusion remains unexplained. Why must a widely used but unsafe device be withdrawn from the market when that particular remedy threatens the health of many and is thus more dangerous than another regulatory response? It is, indeed, a perverse interpretation that reads the FDCA [181] to require the ban of a device that has no "safe" therapeutic purpose where a ban is the most dangerous remedial alternative.

In my view, where linguistically permissible, we should interpret the FDCA in light of Congress' overall desire to protect health. That purpose requires a flexible interpretation that both permits the FDA to take into account the realities of human behavior and allows it, in appropriate cases, to choose from its arsenal of statutory remedies. A statute so interpreted easily "fit[s]" this, and other, drug- and device-related health problems.

In the majority's view, laws enacted since 1965 require us to deny jurisdiction, whatever the FDCA might mean in their absence. But why? Do those laws contain language barring FDA jurisdiction? The majority must concede that they do not. Do they contain provisions that are inconsistent with the FDA's exercise of jurisdiction? With one exception, see infra, at 184-185, the majority points to no such provision. Do they somehow repeal the principles of law (discussed in Part II, supra ) that otherwise would lead to the conclusion that the FDA has jurisdiction in this area? The companies themselves deny making any such claim. See Tr. of Oral Arg. 27 (denying reliance on doctrine of "partial repeal"). Perhaps the later laws "shape" and "focus" what the 1938 Congress meant a generation earlier. Ante, at 143. But this Court has warned against using the views of a later Congress to construe a statute enacted many years before. See Pension Benefit Guaranty Corporation v. LTV Corp., 496 U. S. 633, 650 (1990) (later history is a "`hazardous basis for inferring the intent of an earlier' Congress" (quoting United States v. Price, 361 U. S. 304, 313 (1960))). And, while the majority suggests that the subsequent history "control[s] our construction" of the FDCA, see ante, at 143 (citation and internal quotation marks omitted), this Court [182] expressly has held that such subsequent views are not "controlling." Haynes v. United States, 390 U. S. 85, 87-88, n. 4 (1968); accord, Southwestern Cable Co., 392 U. S., at 170 (such views have "`very little, if any, significance' "); see also Sullivan v. Finkelstein, 496 U. S. 617, 632 (1990) (Scalia, J., concurring) ("Arguments based on subsequent legislative history . . . should not be taken seriously, not even in a footnote").

Regardless, the later statutes do not support the majority's conclusion. That is because, whatever individual Members of Congress after 1964 may have assumed about the FDA's jurisdiction, the laws they enacted did not embody any such "no jurisdiction" assumption. And one cannot automatically infer an antijurisdiction intent, as the majority does, for the later statutes are both (and similarly) consistent with quite a different congressional desire, namely, the intent to proceed without interfering with whatever authority the FDA otherwise may have possessed. See, e. g., Cigarette Labeling and Advertising—1965: Hearings on H. R. 2248 et al. before the House Committee on Interstate and Foreign Commerce, 89th Cong., 1st Sess., 19 (1965) (hereinafter 1965 Hearings) (statement of Rep. Fino that the proposed legislation would not "erode" agency authority). As I demonstrate below, the subsequent legislative history is critically ambivalent, for it can be read either as (a) "ratif[ying]" a no-jurisdiction assumption, see ante, at 158, or as (b) leaving the jurisdictional question just where Congress found it. And the fact that both inferences are "equally tenable," Pension Benefit Guaranty Corp., supra, at 650 (citation and internal quotation marks omitted); Johnson v. Transportation Agency, Santa Clara Cty., 480 U. S. 616, 672 (1987) (Scalia, J., dissenting), prevents the majority from drawing from the later statutes the firm, antijurisdiction implication that it needs.

Consider, for example, Congress' failure to provide the FDA with express authority to regulate tobacco—a circumstance [183] that the majority finds significant. See ante, at 144, 147-148, 155. But cf. Southwestern Cable Co., supra, at 170 (failed requests do not prove agency "did not already possess" authority). In fact, Congress both failed to grant express authority to the FDA when the FDA denied it had jurisdiction over tobacco and failed to take that authority expressly away when the agency later asserted jurisdiction. See, e. g., S. 1262, 104th Cong., 1st Sess., § 906 (1995) (failed bill seeking to amend FDCA to say that "[n]othing in this Act or any other Act shall provide the [FDA] with any authority to regulate in any manner tobacco or tobacco products"); see also H. R. 516, 105th Cong., 1st Sess., § 2 (1997) (similar); H. R. Res. 980, reprinted in 142 Cong. Rec. 5018 (1996) (Georgia legislators unsuccessfully requested that Congress "rescind any action giving the FDA authority" over tobacco); H. R. 2283, 104th Cong., 1st Sess. (1995) (failed bill "[t]o prohibit the [FDA] regulation of the sale or use of tobacco"); H. R. 2414, 104th Cong., 1st Sess., § 2(a) (1995) (similar). Consequently, the defeat of various different proposed jurisdictional changes proves nothing. This history shows only that Congress could not muster the votes necessary either to grant or to deny the FDA the relevant authority. It neither favors nor disfavors the majority's position.

The majority also mentions the speed with which Congress acted to take jurisdiction away from other agencies once they tried to assert it. See ante, at 145, 149-151. But such a congressional response again proves nothing. On the one hand, the speedy reply might suggest that Congress somehow resented agency assertions of jurisdiction in an area it desired to reserve for itself—a consideration that supports the majority. On the other hand, Congress' quick reaction with respect to other agencies' regulatory efforts contrasts dramatically with its failure to enact any responsive law (at any speed) after the FDA asserted jurisdiction over tobacco more than three years ago. And that contrast supports the opposite conclusion.

[184] In addition, at least one post-1938 statute reveals quite a different congressional intent than the majority infers. See note following 21 U. S. C. § 321 (1994 ed., Supp. III) (FDA Modernization Act of 1997) (law "shall [not] be construed to affect the question of whether the [FDA] has any authority to regulate any tobacco product," and "[s]uch authority, if any, shall be exercised under the [FDCA] as in effect on the day before the date of [this] enactment"). Consequently, it appears that the only interpretation that can reconcile all of the subsequent statutes is the inference that Congress did not intend, either explicitly or implicitly, for its later laws to answer the question of the scope of the FDA's jurisdictional authority. See 143 Cong. Rec. S8860 (Sept. 5, 1997) (the Modernization Act will "not interfere or substantially negatively affect any of the FDA tobacco authority").

The majority's historical perspective also appears to be shaped by language in the Federal Cigarette Labeling and Advertising Act (FCLAA), 79 Stat. 282, 15 U. S. C. § 1331 et seq. See ante, at 148-149. The FCLAA requires manufacturers to place on cigarette packages, etc., health warnings such as the following:

"SURGEON GENERAL'S WARNING: Smoking Causes Lung Cancer, Heart Disease, Emphysema, And May Complicate Pregnancy." 15 U. S. C. § 1333(a).

The FCLAA has an express pre-emption provision which says that "[n]o statement relating to smoking and health, other than the statement required by [this Act], shall be required on any cigarette package." § 1334(a). This preemption clause plainly prohibits the FDA from requiring on "any cigarette package" any other "statement relating to smoking and health," but no one contends that the FDA has failed to abide by this prohibition. See, e. g., 61 Fed. Reg. 44399 (1996) (describing the other regulatory prescriptions). Rather, the question is whether the FCLAA's pre-emption [185] provision does more. Does it forbid the FDA to regulate at all?

This Court has already answered that question expressly and in the negative. See Cipollone v. Liggett Group, Inc., 505 U. S. 504 (1992). Cipollone held that the FCLAA's preemption provision does not bar state or federal regulation outside the provision's literal scope. Id., at 518. And it described the pre-emption provision as "merely prohibit[ing] state and federal rulemaking bodies from mandating particular cautionary statements on cigarette labels . . . ." Ibid.

This negative answer is fully consistent with Congress' intentions in regard to the pre-emption language. When Congress enacted the FCLAA, it focused upon the regulatory efforts of the Federal Trade Commission (FTC), not the FDA. See 1965 Hearings 1-2. And the Public Health Cigarette Smoking Act of 1969, Pub. L. 91-222, § 7(c), 84 Stat. 89, expressly amended the FCLAA to provide that "[n]othing in this Act shall be construed to affirm or deny the [FTC's] holding that it has the authority to issue trade regulation rules" for tobacco. See also H. R. Conf. Rep. No. 91-897, p. 7 (1970) (statement of House Managers) (we have "no intention to resolve the question as to whether" the FTC could regulate tobacco in a different way); see also 116 Cong. Rec. 7921 (1970) (statement of Rep. Satterfield) (same). Why would one read the FCLAA's pre-emption clause—a provision that Congress intended to limit even in respect to the agency directly at issue—so broadly that it would bar a different agency from engaging in any other cigarette regulation at all? The answer is that the Court need not, and should not, do so. And, inasmuch as the Court already has declined to view the FCLAA as pre-empting the entire field of tobacco regulation, I cannot accept that that same law bars the FDA's regulatory efforts here.

When the FCLAA's narrow pre-emption provision is set aside, the majority's conclusion that Congress clearly intended for its tobacco-related statutes to be the exclusive [186] "response" to "the problem of tobacco and health," ante, at 157, is based on legislative silence. Notwithstanding the views voiced by various legislators, Congress itself has addressed expressly the issue of the FDA's tobacco-related authority only once—and, as I have said, its statement was that the statute was not to "be construed to affect the question of whether the [FDA] has any authority to regulate any tobacco product." Note following 21 U. S. C. § 321 (1994 ed., Supp. III). The proper inference to be drawn from all of the post-1965 statutes, then, is one that interprets Congress' general legislative silence consistently with this statement.

I now turn to the final historical fact that the majority views as a factor in its interpretation of the subsequent legislative history: the FDA's former denials of its tobaccorelated authority.

Until the early 1990's, the FDA expressly maintained that the 1938 statute did not give it the power that it now seeks to assert. It then changed its mind. The majority agrees with me that the FDA's change of positions does not make a significant legal difference. See ante, at 156-157; see also Chevron, 467 U. S., at 863 ("An initial agency interpretation is not instantly carved in stone"); accord, Smiley v. Citibank (South Dakota), N. A., 517 U. S. 735, 742 (1996) ("[C]hange is not invalidating"). Nevertheless, it labels those denials "important context" for drawing an inference about Congress' intent. Ante, at 157. In my view, the FDA's change of policy, like the subsequent statutes themselves, does nothing to advance the majority's position.

When it denied jurisdiction to regulate cigarettes, the FDA consistently stated why that was so. In 1963, for example, FDA administrators wrote that cigarettes did not satisfy the relevant FDCA definitions—in particular, the "intent" requirement—because cigarette makers did not sell their product with accompanying "therapeutic claims." [187] Letter to Directors of Bureaus, Divisions and Directors of Districts from FDA Bureau of Enforcement (May 24, 1963), in Public Health Cigarette Amendments of 1971: Hearings on S. 1454 before the Consumer Subcommittee of the Senate Committee on Commerce, 92d Cong., 2d Sess., 240 (1972) (hereinafter FDA Enforcement Letter). And subsequent FDA Commissioners made roughly the same assertion. One pointed to the fact that the manufacturers only "recommended" cigarettes "for smoking pleasure." Two others reiterated the evidentiary need for "health claims." Yet another stressed the importance of proving "intent," adding that "[w]e have not had sufficient evidence" of "intent with regard to nicotine." See, respectively, id., at 239 (Comm'r Edwards); Letter of Dec. 5, 1977, App. 47 (Comm'r Kennedy); 1965 Hearings 193 (Comm'r Rankin); 1994 Hearings 28 (Comm'r Kessler). Tobacco company counsel also testified that the FDA lacked jurisdiction because jurisdiction "depends on . . . intended use," which in turn "depends, in general, on the claims and representations made by the manufacturer." Health Consequences of Smoking: Nicotine Addiction, Hearing before the Subcommittee on Health and the Environment of the House Committee on Energy and Commerce, 100th Cong., 2d Sess., 288 (1988) (testimony of Richard Cooper) (emphasis added).

Other agency statements occasionally referred to additional problems. Commissioner Kessler, for example, said that the "enormous social consequences" flowing from a decision to regulate tobacco counseled in favor of obtaining specific congressional "guidance." 1994 Hearings 69; see also ante, at 153 (quoting statement of Health and Human Services Secretary Brandt to the effect that Congress wanted to make the relevant jurisdictional decision). But a fair reading of the FDA's denials suggests that the overwhelming problem was one of proving the requisite manufacturer intent. See Action on Smoking and Health v. Harris, 655 F. 2d 236, 238-239 (CADC 1980) (FDA "comments" reveal its "understanding" [188] that "the crux of FDA jurisdiction over drugs lay in manufacturers' representations as revelatory of their intent").

What changed? For one thing, the FDA obtained evidence sufficient to prove the necessary "intent" despite the absence of specific "claims." See supra, at 172-174. This evidence, which first became available in the early 1990's, permitted the agency to demonstrate that the tobacco companies knew nicotine achieved appetite-suppressing, mood-stabilizing, and habituating effects through chemical (not psychological) means, even at a time when the companies were publicly denying such knowledge.

Moreover, scientific evidence of adverse health effects mounted, until, in the late 1980's, a consensus on the seriousness of the matter became firm. That is not to say that concern about smoking's adverse health effects is a new phenomenon. See, e. g., Higginson, A New Counterblast, in Out-door Papers 179, 194 (1863) (characterizing tobacco as "`a narcotic poison of the most active class' "). It is to say, however, that convincing epidemiological evidence began to appear mid-20th century; that the first Surgeon General's Report documenting the adverse health effects appeared in 1964; and that the Surgeon General's Report establishing nicotine's addictive effects appeared in 1988. At each stage, the health conclusions were the subject of controversy, diminishing somewhat over time, until recently—and only recently—has it become clear that there is a wide consensus about the health problem. See 61 Fed. Reg. 44701-44706 (1996).

Finally, administration policy changed. Earlier administrations may have hesitated to assert jurisdiction for the reasons prior Commissioners expressed. See supra, at 186-187 and this page. Commissioners of the current administration simply took a different regulatory attitude.

Nothing in the law prevents the FDA from changing its policy for such reasons. By the mid-1990's, the evidence [189] needed to prove objective intent—even without an express claim—had been found. The emerging scientific consensus about tobacco's adverse, chemically induced, health effects may have convinced the agency that it should spend its resources on this important regulatory effort. As for the change of administrations, I agree with then-Justice Rehnquist's statement in a different case, where he wrote:

"The agency's changed view . . . seems to be related to the election of a new President of a different political party. It is readily apparent that the responsible members of one administration may consider public resistance and uncertainties to be more important than do their counterparts in a previous administration. A change in administration brought about by the people casting their votes is a perfectly reasonable basis for an executive agency's reappraisal of the costs and benefits of its programs and regulations. As long as the agency remains within the bounds established by Congress, it is entitled to assess administrative records and evaluate priorities in light of the philosophy of the administration." Motor Vehicle Mfrs. Assn. of United States, Inc. v. State Farm Mut. Automobile Ins. Co., 463 U. S. 29, 59 (1983) (concurring in part and dissenting in part).

One might nonetheless claim that, even if my interpretation of the FDCA and later statutes gets the words right, it lacks a sense of their "music." See Helvering v. Gregory, 69 F. 2d 809, 810-811 (CA2 1934) (L. Hand, J.) ("[T]he meaning of a [statute] may be more than that of the separate words, as a melody is more than the notes . . .").Such a claim might rest on either of two grounds.

First, one might claim that, despite the FDA's legal right to change its mind, its original statements played a critical part in the enactment of the later statutes and now should play a critical part in their interpretation. But the FDA's [190] traditional view was largely premised on a perceived inability to prove the necessary statutory "intent" requirement. See, e. g., FDA Enforcement Letter 240 ("The statutory basis for the exclusion of tobacco products from FDA's jurisdiction is the fact that tobacco marketed for chewing or smoking without accompanying therapeutic claims, does not meet the definitions . . . for food, drug, device or cosmetic"). The statement, "we cannot assert jurisdiction over substance X unless it is treated as a food," would not bar jurisdiction if the agency later establishes that substance X is, and is intended to be, eaten. The FDA's denials of tobacco-related authority sufficiently resemble this kind of statement that they should not make the critical interpretive difference.

Second, one might claim that courts, when interpreting statutes, should assume in close cases that a decision with "enormous social consequences," 1994 Hearings 69, should be made by democratically elected Members of Congress rather than by unelected agency administrators. Cf. Kent v. Dulles, 357 U. S. 116, 129 (1958) (assuming Congress did not want to delegate the power to make rules interfering with exercise of basic human liberties). If there is such a background canon of interpretation, however, I do not believe it controls the outcome here.

Insofar as the decision to regulate tobacco reflects the policy of an administration, it is a decision for which that administration, and those politically elected officials who support it, must (and will) take responsibility. And the very importance of the decision taken here, as well as its attendant publicity, means that the public is likely to be aware of it and to hold those officials politically accountable. Presidents, just like Members of Congress, are elected by the public. Indeed, the President and Vice President are the only public officials whom the entire Nation elects. I do not believe that an administrative agency decision of this magnitude—one that is important, conspicuous, and controversial—can escape the kind of public scrutiny that is essential in any democracy. [191] And such a review will take place whether it is the Congress or the Executive Branch that makes the relevant decision.

* * *

According to the FDA, only 2.5% of smokers successfully stop smoking each year, even though 70% say they want to quit and 34% actually make an attempt to do so. See 61 Fed. Reg. 44704 (1996) (citing Centers for Disease Control and Prevention, Cigarette Smoking Among Adults—United States, 1993; 43 Morbidity and Mortality Weekly Report 929 (Dec. 23, 1994)). The fact that only a handful of those who try to quit smoking actually succeed illustrates a certain reality—the reality that the nicotine in cigarettes creates a powerful physiological addiction flowing from chemically induced changes in the brain. The FDA has found that the makers of cigarettes "intend" these physical effects. Hence, nicotine is a "drug"; the cigarette that delivers nicotine to the body is a "device"; and the FDCA's language, read in light of its basic purpose, permits the FDA to assert the disease-preventing jurisdiction that the agency now claims.

The majority finds that cigarettes are so dangerous that the FDCA would require them to be banned (a result the majority believes Congress would not have desired); thus, it concludes that the FDA has no tobacco-related authority. I disagree that the statute would require a cigarette ban. But even if I am wrong about the ban, the statute would restrict only the agency's choice of remedies, not its jurisdiction.

The majority also believes that subsequently enacted statutes deprive the FDA of jurisdiction. But the later laws say next to nothing about the FDA's tobacco-related authority. Previous FDA disclaimers of jurisdiction may have helped to form the legislative atmosphere out of which Congress' own tobacco-specific statutes emerged. But a legislative atmosphere is not a law, unless it is embodied in a statutory word or phrase. And the relevant words and phrases here reveal [192] nothing more than an intent not to change the jurisdictional status quo.

The upshot is that the Court today holds that a regulatory statute aimed at unsafe drugs and devices does not authorize regulation of a drug (nicotine) and a device (a cigarette) that the Court itself finds unsafe. Far more than most, this particular drug and device risks the life-threatening harms that administrative regulation seeks to rectify. The majority's conclusion is counterintuitive. And, for the reasons set forth, I believe that the law does not require it.

Consequently, I dissent.

[1] Briefs of amici curiae urging reversal were filed for the State of Minnesota et al. by Mike Hatch, Attorney General of Minnesota, James S. Alexander, Assistant Attorney General, Louise H. Renne, and by the Attorneys General for their respective States as follows: Bruce M. Botelho of Alaska, Janet Napolitano of Arizona, Mark Pryor of Arkansas, Bill Lockyer of California, Ken Salazar of Colorado, Richard Blumenthal of Connecticut, Robert A. Butterworth of Florida, Earl I. Anzai of Hawaii, Alan G. Lance of Idaho, James E. Ryan of Illinois, Jeffrey A. Modisett of Indiana, Thomas J. Miller of Iowa, Carla J. Stovall of Kansas, Andrew Ketterer of Maine, J. Joseph Curran, Jr., of Maryland, Thomas F. Reilly of Massachusetts, Jennifer M. Granholm of Michigan, Mike Moore of Mississippi, Jeremiah W. Nixon of Missouri, Joseph P. Mazurek of Montana, Frankie Sue Del Papa of Nevada, Philip T. McLaughlin of New Hampshire, John J. Farmer, Jr., of New Jersey, Patricia A. Madrid of New Mexico, Eliot Spitzer of New York, Heidi Heitkamp of North Dakota, Betty D. Montgomery of Ohio, W. A. Drew Edmondson of Oklahoma, Hardy Myers of Oregon, D. Michael Fisher of Pennsylvania, Sheldon Whitehouse of Rhode Island, Mark Barnett of South Dakota, John Cornyn of Texas, Jan Graham of Utah, William H. Sorrell of Vermont, Christine O. Gregoire of Washington, Darrell V. McGraw, Jr., of West Virginia, James E. Doyle of Wisconsin, and Gay Woodhouse of Wyoming; for Action on Smoking and Health by John F. Banzhaf III and Kathleen E. Scheg; for the American Cancer Society, Inc., by Russell E. Brooks, David R. Gelfand, Charles W. Westland, and William J. Dalton; for the American College of Chest Physicians by Raymond D. Cotton; and for Public Citizen, Inc., et al. by Allison M. Zieve, Alan B. Morrison, and David C. Vladeck.

Briefs of amici curiae urging affirmance were filed for the Pacific Legal Foundation by Anne M. Hayes and M. Reed Hopper; for the Product Liability Advisory Council, Inc., by Kenneth S. Geller; and for the Washington Legal Foundation et al. by Daniel J. Popeo and Richard A. Samp.