*435Justice White
delivered the opinion of the Court.
A Texas city denied a special use permit for the operation of a group home for the mentally retarded, acting pursuant to a municipal zoning ordinance requiring permits for such homes. The Court of Appeals for the Fifth Circuit held that mental retardation is a “quasi-suspect” classification and that the ordinance violated the Equal Protection Clause because it did not substantially further an important governmental purpose. We hold that a lesser standard of scrutiny is appropriate, but conclude that under that standard the ordinance is invalid as applied in this case.
h — I
In July 1980, respondent Jan Hannah purchased a building at 201 Featherston Street in the city of Cleburne, Texas, with the intention of leasing it to Cleburne Living Center, Inc. (CLC), for the operation of a group home for the mentally retarded. It was anticipated that the home would house 13 retarded men and women, who would be under the constant supervision of CLC staff members. The house had four bedrooms and two baths, with a half bath to be added. CLC planned to comply with all applicable state and federal regulations.
*436The city informed CLC that a special use permit would be required for the operation of a group home at the site, and CLC accordingly submitted a permit application. In response to a subsequent inquiry from CLC, the city explained that under the zoning regulations applicable to the site, a special use permit, renewable annually, was required for the construction of “[hjospitals for the insane or feeble-minded, or alcoholic [sic] or drug addicts, or penal or correctional institutions.” The city had determined that the proposed *437group home should be classified as a “hospital for the feeble-minded.” After holding a public hearing on CLC’s application, the City Council voted 3 to 1 to deny a special use permit.
CLC then filed suit in Federal District Court against the city and a number of its officials, alleging, inter alia, that the zoning ordinance was invalid on its face and as applied because it discriminated against the mentally retarded in violation of the equal protection rights of CLC and its potential residents. The District Court found that “[i]f the potential residents of the Featherston Street home were not mentally retarded, but the home was the same in all other respects, its use would be permitted under the city’s zoning ordinance,” and that the City Council’s decision “was motivated primarily by the fact that the residents of the home would be persons who are mentally retarded.” App. 93, 94. Even so, the District Court held the ordinance and its application constitutional. Concluding that no fundamental right was implicated and that mental retardation was neither a suspect nor a quasi-suspect classification, the court employed the minimum level of judicial scrutiny applicable to equal protection claims. The court deemed the ordinance, as written and applied, to be rationally related to the city’s legitimate interests in “the legal responsibility of CLC and its residents, . . . the safety and fears of residents in the adjoining neighborhood,” and the number of people to be housed in the home. Id., at 103.
The Court of Appeals for the Fifth Circuit reversed, determining that mental retardation was a quasi-suspect classification and that it should assess the validity of the ordinance *438under intermediate-level scrutiny. 726 F. 2d 191 (1984). Because mental retardation was in fact relevant to many legislative actions, strict scrutiny was not appropriate. But in light of the history of “unfair and often grotesque mistreatment” of the retarded, discrimination against them was “likely to reflect deep-seated prejudice.” Id., at 197. In addition, the mentally retarded lacked political power, and their condition was immutable. The court considered heightened scrutiny to be particularly appropriate in this case, because the city’s ordinance withheld a benefit which, although not fundamental, was very important to the mentally retarded. Without group homes, the court stated, the retarded could never hope to integrate themselves into the community. Applying the test that it considered appropriate, the court held that the ordinance was invalid on its face because it did not substantially further any important governmental interests. The Court of Appeals went on to hold that the ordinance was also invalid as applied. Rehearing en banc was *439denied with six judges dissenting in an opinion urging en banc consideration of the panel’s adoption of a heightened standard of review. We granted certiorari, 469 U. S. 1016 (1984).
1 — I t — i
The Equal Protection Clause of the Fourteenth Amendment commands that no State shall “deny to any person within its jurisdiction the equal protection of the laws,” which is essentially a direction that all persons similarly situated should be treated alike. Plyler v. Doe, 457 U. S. 202, 216 (1982). Section 5 of the Amendment empowers Congress to enforce this mandate, but absent controlling congressional direction, the courts have themselves devised standards for *440determining the validity of state legislation or other official action that is challenged as denying equal protection. The general rule is that legislation is presumed to be valid and will be sustained if the classification drawn by the statute is rationally related to a legitimate state interest. Schweiker v. Wilson, 450 U. S. 221, 230 (1981); United States Railroad Retirement Board v. Fritz, 449 U. S. 166, 174-175 (1980); Vance v. Bradley, 440 U. S. 93, 97 (1979); New Orleans v. Dukes, 427 U. S. 297, 303 (1976). When social or economic legislation is at issue, the Equal Protection Clause allows the States wide latitude, United States Railroad Retirement Board v. Fritz, supra, at 174; New Orleans v. Dukes, supra, at 303, and the Constitution presumes that even improvident decisions will eventually be rectified by the democratic processes.
The general rule gives way, however, when a statute classifies by race, alienage, or national origin. These factors are so seldom relevant to the achievement of any legitimate state interest that laws grounded in such considerations are deemed to reflect prejudice and antipathy — a view that those in the burdened class are not as worthy or deserving as others. For these reasons and because such discrimination is unlikely to be soon rectified by legislative means, these laws are subjected to strict scrutiny and will be sustained only if they are suitably tailored to serve a compelling state interest. McLaughlin v. Florida, 379 U. S. 184, 192 (1964); Graham v. Richardson, 403 U. S. 365 (1971). Similar oversight by the courts is due when state laws impinge on personal rights protected by the Constitution. Kramer v. Union Free School District No. 15, 395 U. S. 621 (1969); Shapiro v. Thompson, 394 U. S. 618 (1969); Skinner v. Oklahoma ex rel. Williamson, 316 U. S. 535 (1942).
Legislative classifications based on gender also call for a heightened standard of review. That factor generally provides no sensible ground for differential treatment. “[W]hat differentiates sex from such nonsuspect statuses as intelligence or physical disability ... is that the sex characteristic *441frequently bears no relation to ability to perform or contribute to society.” Frontiero v. Richardson, 411 U. S. 677, 686 (1973) (plurality opinion). Rather than resting on meaningful considerations, statutes distributing benefits and burdens between the sexes in different ways very likely reflect outmoded notions of the relative capabilities of men and women. A gender classification fails unless it is substantially related to a sufficiently important governmental interest. Mississippi University for Women v. Hogan, 458 U. S. 718 (1982); Craig v. Boren, 429 U. S. 190 (1976). Because illegitimacy is beyond the individual’s control and bears “no relation to the individual’s ability to participate in and contribute to society,” Mathews v. Lucas, 427 U. S. 495, 505 (1976), official discriminations resting on that characteristic are also subject to somewhat heightened review. Those restrictions “will survive equal protection scrutiny to the extent they are substantially related to a legitimate state interest.” Mills v. Habluetzel, 456 U. S. 91, 99 (1982).
We have declined, however, to extend heightened review to differential treatment based on age:
“While the treatment of the aged in this Nation has not been wholly free of discrimination, such persons, unlike, say, those who have been discriminated against on the basis of race or national origin, have not experienced a ‘history of purposeful unequal treatment’ or been subjected to unique disabilities on the basis of stereotyped characteristics not truly indicative of their abilities.” Massachusetts Board of Retirement v. Murgia, 427 U. S. 307, 313 (1976).
The lesson of Murgia is that where individuals in the group affected by a law have distinguishing characteristics relevant to interests the State has the authority to implement, the courts have been very reluctant, as they should be in our federal system and with our respect for the separation of powers, to closely scrutinize legislative choices as to whether, how, and to what extent those interests should be *442pursued. In such cases, the Equal Protection Clause requires only a rational means to serve a legitimate end.
I — I hH h-H
Against this background, we conclude for several reasons that the Court of Appeals erred in holding mental retardation a quasi-suspect classification calling for a more exacting standard of judicial review than is normally accorded economic and social legislation. First, it is undeniable, and it is not argued otherwise here, that those who are mentally retarded have a reduced ability to cope with and function in the everyday world. Nor are they all cut from the same pattern: as the testimony in this record indicates, they range from those whose disability is not immediately evident to those who must be constantly cared for. They are thus different, immutably so, in relevant respects, and the States’ interest in dealing with and providing for them is plainly a legitimate one. How this large and diversified group is to be treated *443under the law is a difficult and often a technical matter, very much a task for legislators guided by qualified professionals and not by the perhaps ill-informed opinions of the judiciary. Heightened scrutiny inevitably involves substantive judgments about legislative decisions, and we doubt that the predicate for such judicial oversight is present where the classification deals with mental retardation.
Second, the distinctive 'legislative response, both national and state, to the plight of those who are mentally retarded demonstrates not only that they have unique problems, but also that the lawmakers have been addressing their difficulties in a manner that belies a continuing antipathy or prejudice and a corresponding need for more intrusive oversight by the judiciary. Thus, the Federal Government has not only outlawed discrimination against the mentally retarded in federally funded programs, see § 504 of the Rehabilitation Act of 1973, 29 U. S. C. § 794, but it has also provided the retarded with the right to receive “appropriate treatment, services, and habilitation” in a setting that is “least restrictive of [their] personal liberty.” Developmental Disabilities Assistance and Bill of Rights Act, 42 U. S. C. §§6010(1), (2). In addition, the Government has conditioned federal education funds on a State’s assurance that retarded children will enjoy an education that, “to the maximum extent appropriate,” is integrated with that of nonmentally retarded children. Education of the Handicapped Act, 20 U. S. C. § 1412(5)(B). The Government has also facilitated the hiring of the mentally retarded into the federal civil service by exempting them from the requirement of competitive examina*444tion. See 5 CFR §213.3102(t) (1984). The State of Texas has similarly enacted legislation that acknowledges the special status of the mentally retarded by conferring certain rights upon them, such as “the right to live in the least restrictive setting appropriate to [their] individual needs and abilities,” including “the right to live ... in a group home.” Mentally Retarded Persons Act of 1977, Tex. Rev. Civ. Stat. Ann., Art. 5547-300, §7 (Vernon Supp. 1985).
Such legislation thus singling out the retarded for special treatment reflects the real and undeniable differences between the retarded and others. That a civilized and decent society expects and approves such legislation indicates that governmental consideration of those differences in the vast majority of situations is not only legitimate but also desirable. It may be, as CLC contends, that legislation designed to benefit, rather than disadvantage, the retarded would generally withstand examination under a test of heightened scrutiny. See Brief for Respondents 38-41. The relevant inquiry, however, is whether heightened scrutiny is constitutionally mandated in the first instance. Even assuming that many of these laws could be shown to be substantially related to an important governmental purpose, merely requiring the legislature to justify its efforts in these terms may lead it to refrain from acting at all. Much recent legislation intended to benefit the retarded also assumes the need for measures that might be perceived to disadvantage them. The Education of the Handicapped Act, for example, requires an “appropriate” education, not one that is equal in all respects *445to the education of nonretarded children; clearly, admission to a class that exceeded the abilities of a retarded child would not be appropriate. Similarly, the Developmental Disabilities Assistance Act and the Texas Act give the retarded the right to live only in the “least restrictive setting” appropriate to their abilities, implicitly assuming the need for at least some restrictions that would not be imposed on others. Especially given the wide variation in the abilities and needs of the retarded themselves, governmental bodies must have a certain amount of flexibility and freedom from judicial oversight in shaping and limiting their remedial efforts.
Third, the legislative response, which could hardly have occurred and survived without public support, negates any claim that the mentally retarded are politically powerless in the sense that they have no ability to attract the attention of the lawmakers. Any minority can be said to be powerless to assert direct control over the legislature, but if that were a criterion for higher level scrutiny by the courts, much economic and social legislation would now be suspect.
Fourth, if the large and amorphous class of the mentally retarded were deemed quasi-suspect for the reasons given by the Court of Appeals, it would be difficult to find a principled way to distinguish a variety of other groups who have perhaps immutable disabilities setting them off from others, who cannot themselves mandate the desired legislative responses, and who can claim some degree of prejudice from at least part of the public at large. One need mention in this respect only *446the aging, the disabled, the mentally ill, and the infirm. We are reluctant to set out on that course, and we decline to do so.
Doubtless, there have been and there will continue to be instances of discrimination against the retarded that are in fact invidious, and that are properly subject to judicial correction under constitutional norms. But the appropriate method of reaching such instances is not to create a new quasi-suspect classification and subject all governmental action based on that classification to more searching evaluation. Rather, we should look to the likelihood that governmental action premised on a particular classification is valid as a general matter, not merely to the specifics of the case before us. Because mental retardation is a characteristic that the government may legitimately take into account in a wide range of decisions, and because both State and Federal Governments have recently committed themselves to assisting the retarded, we will not presume that any given legislative action, even one that disadvantages retarded individuals, is rooted in considerations that the Constitution will not tolerate.
Our refusal to recognize the retarded as a quasi-suspect class does not leave them entirely unprotected from invidious discrimination. To withstand equal protection review, legislation that distinguishes between the mentally retarded and others must be rationally related to a legitimate governmental purpose. This standard, we believe, affords government the latitude necessary both to pursue policies designed to assist the retarded in realizing their full potential, and to freely and efficiently engage in activities that burden the retarded in what is essentially an incidental manner. The State may not rely on a classification whose relationship to an asserted goal is so attenuated as to render the distinction arbitrary or irrational. See Zobel v. Williams, 457 U. S. 55, 61-63 (1982); United States Dept. of Agriculture v. Moreno, 413 U. S. 528, 535 (1973). Furthermore, some objectives — *447such as “a bare . . . desire to harm a politically unpopular group,” id,., at 534 — are not legitimate state interests. See also Zobel, supra, at 63. Beyond that, the mentally retarded, like others, have and retain their substantive constitutional rights in addition to the right to be treated equally by the law.
IV
We turn to the issue of the validity of the zoning ordinance insofar as it requires a special use permit for homes for the mentally retarded. We inquire first whether requiring a special use permit for the Featherston home in the circumstances here deprives respondents of the equal protection of the laws. If it does, there will be no occasion to decide whether the special use permit provision is facially invalid where the mentally retarded are involved, or to put it another way, whether the city may never insist on a special use permit for a home for the mentally retarded in an R-3 zone. This is the preferred course of adjudication since it enables courts to avoid making unnecessarily broad constitutional judgments. Brockett v. Spokane Arcades, Inc., 472 U. S. 491, 501-502 (1985); United States v. Grace, 461 U. S. 171 (1983); NAACP v. Button, 371 U. S. 415 (1963).
The constitutional issue is clearly posed. The city does not require a special use permit in an R-3 zone for apartment houses, multiple dwellings, boarding and lodging houses, fraternity or sorority houses, dormitories, apartment hotels, hospitals, sanitariums, nursing homes for convalescents or the aged (other than for the insane or feebleminded or alcoholics or drug addicts), private clubs or fraternal orders, and other specified uses. It does, however, insist on a special permit for the Featherston home, and it does so, as the District Court found, because it would be a facility for the men*448tally retarded. May the city require the permit for this facility when other care and multiple-dwelling facilities are freely permitted?
It is true, as already pointed out, that the mentally retarded as a group are indeed different from others not sharing their misfortune, and in this respect they may be different from those who would occupy other facilities that would be permitted in an R-3 zone without a special permit. But this difference is largely irrelevant unless the Featherston home and those who would occupy it would threaten legitimate interests of the city in a way that other permitted uses such as boarding houses and hospitals would not. Because in our view the record does not reveal any rational basis for believing that the Featherston home would pose any special threat to the city’s legitimate interests, we affirm the judgment below insofar as it holds the ordinance invalid as applied in this case.
The District Court found that the City Council’s insistence on the permit rested on several factors. First, the Council was concerned with the negative attitude of the majority of property owners located within 200 feet of the Featherston facility, as well as with the fears of elderly residents of the neighborhood. But mere negative attitudes, or fear, unsubstantiated by factors which are properly cognizable in a zoning proceeding, are not permissible bases for treating a home for the mentally retarded differently from apartment houses, multiple dwellings, and the like. It is plain that the electorate as a whole, whether by referendum or otherwise, could not order city action violative of the Equal Protection Clause, Lucas v. Forty-Fourth General Assembly of Colorado, 377 U. S. 713, 736-737 (1964), and the city may not avoid the strictures of that Clause by deferring to the wishes or objections of some fraction of the body politic. “Private biases may be outside the reach of the law, but the law cannot, directly or indirectly, give them effect.” Palmore v. Sidoti, 466 U. S. 429, 433 (1984).
*449Second, the Council had two objections to the location of the facility. It was concerned that the facility was across the street from a junior high school, and it feared that the students might harass the occupants of the Featherston home. But the school itself is attended by about 30 mentally retarded students, and denying a permit based on such vague, undifferentiated fears is again permitting some portion of the community to validate what would otherwise be an equal protection violation. The other objection to the home’s location was that it was located on “a five hundred year flood plain.” This concern with the possibility of a flood, however, can hardly be based on a distinction between the Featherston home and, for example, nursing homes, homes for convalescents or the aged, or sanitariums or hospitals, any of which could be located on the Featherston site without obtaining a special use permit. The same may be said of another concern of the Council — doubts about the legal responsibility for actions which the mentally retarded might take. If there is no concern about legal responsibility with respect to other uses that would be.permitted in the area, such as boarding and fraternity houses, it is difficult to believe that the groups of mildly or moderately mentally retarded individuals who would live at 201 Featherston would present any different or special hazard.
Fourth, the Council was concerned with the size of the home and the number of people that would occupy it. The District Court found, and the Court of Appeals repeated, that “[i]f the potential residents of the Featherston Street home were not mentally retarded, but the home was the same in all other respects, its use would be permitted under the city’s zoning ordinance.” App. 93; 726 F. 2d, at 200. Given this finding, there would be no restrictions on the number of people who could occupy this home as a boarding house, nursing home, family dwelling, fraternity house, or dormitory. The question is whether it is rational to treat the mentally retarded differently. It is true that they suffer dis*450ability not shared by others; but why this difference warrants a density regulation that others need not observe is not at all apparent. At least this record does not clarify how, in this connection, the characteristics of the intended occupants of the Featherston home rationally justify denying to those occupants what would be permitted to groups occupying the same site for different purposes. Those who would live in the Featherston home are the type of individuals who, with supporting staff, satisfy federal and state standards for group housing in the community; and there is no dispute that the home would meet the federal square-footage-per-resident requirement for facilities of this type. See 42 CFR § 442.447 (1984). In the words of the Court of Appeals, “[t]he City never justifies its apparent view that other people can live under such ‘crowded’ conditions when mentally retarded persons cannot.” 726 F. 2d, at 202.
In the courts below the city also urged that the ordinance is aimed at avoiding concentration of population and at lessening congestion of the streets. These concerns obviously fail to explain why apartment houses, fraternity and sorority houses, hospitals and the like, may freely locate in the area without a permit. So, too, the expressed worry about fire hazards, the serenity of the neighborhood, and the avoidance of danger to other residents fail rationally to justify singling out a home such as 201 Featherston for the special use permit, yet imposing no such restrictions on the many other uses freely permitted in the neighborhood.
The short of it is that requiring the permit in this case appears to us to rest on an irrational prejudice against the mentally retarded, including those who would occupy the-Featherston facility and who would live under the closely supervised and highly regulated conditions expressly provided for by state and federal law.
The judgment of the Court of Appeals is affirmed insofar as it invalidates the zoning ordinance as applied to the Featherston home. The judgment is otherwise vacated, and the case is remanded.
It is so ordered.
Justice Marshall,
with whom Justice Brennan and Justice Blackmun join, concurring in the judgment in part and dissenting in part.
The Court holds that all retarded individuals cannot be grouped together as the “feebleminded” and deemed presumptively unfit to live in a community. Underlying this holding is the principle that mental retardation per se cannot be a proxy for depriving retarded people of their rights and interests without regard to variations in individual ability. *456With this holding and principle I agree. The Equal Protection Clause requires attention to the capacities and needs of retarded people as individuals.
I cannot agree, however, with the way in which the Court reaches its result or with the narrow, as-applied remedy it provides for the city of Cleburne’s equal protection violation. The Court holds the ordinance invalid on rational-basis grounds and disclaims that anything special, in the form of heightened scrutiny, is taking place. Yet Cleburne’s ordinance surely would be valid under the traditional rational-basis test applicable to economic and commercial regulation. In my view, it is important to articulate, as the Court does not, the facts and principles that justify subjecting this zoning ordinance to the searching review — the heightened scrutiny — that actually leads to its invalidation. Moreover, in invalidating Cleburne’s exclusion of the “feebleminded” only as applied to respondents, rather than on its face, the Court radically departs from our equal protection precedents. Because I dissent from this novel and truncated remedy, and because I cannot accept the Court’s disclaimer that no “more exacting standard” than ordinary rational-basis review is being applied, ante, at 442,1 write separately.
I
At the outset, two curious and paradoxical aspects of the Court’s opinion must be noted. First, because the Court invalidates Cleburne’s zoning ordinance on rational-basis grounds, the Court’s wide-ranging discussion of heightened scrutiny is wholly superfluous to the decision of this case. This “two for the price of one” approach to constitutional decisionmaking — rendering two constitutional rulings where one is enough to decide the case — stands on their head traditional and deeply embedded principles governing exercise of the Court’s Article III power. Just a few weeks ago, the Court “call[ed] to mind two of the cardinal rules governing *457the federal courts: ‘One, never to anticipate a question of constitutional law in advance of the necessity of deciding it; the other never to formulate a rule of constitutional law broader than is required by the precise facts to which it is to be applied.’” Brockett v. Spokane Arcades, Inc., 472 U. S. 491, 501 (1985) (White, J.) (quoting Liverpool, New York & Philadelphia S.S. Co. v. Commissioners of Emigration, 113 U. S. 33, 39 (1885)). When a lower court correctly, decides a case, albeit on what this Court concludes are unnecessary constitutional grounds, “our usual custom” is not to compound the problem by following suit but rather to affirm on the narrower, dispositive ground available. Alexander v. Louisiana, 405 U. S. 625, 633 (1972). The Court offers no principled justification for departing from these principles, nor, given our equal protection precedents, could it. See Mississippi University for Women v. Hogan, 458 U. S. 718, 724, n. 9 (1982) (declining to address strict scrutiny when heightened scrutiny sufficient to invalidate action challenged); Stanton v. Stanton, 421 U. S. 7, 13 (1975) *458(same); Hooper v. Bernalillo County Assessor, 472 U. S. 612, 618 (1985) (declining to reach heightened scrutiny in review of residency-based classifications that fail rational-basis test); Zobel v. Williams, 457 U. S. 55, 60-61 (1982) (same); cf. Mitchell v. Forsyth, 472 U. S. 511, 537-538 (1985) (O’Connor, J., concurring in part).
Second, the Court’s heightened-scrutiny discussion is even more puzzling given that Cleburne’s ordinance is invalidated only after being subjected to precisely the sort of probing inquiry associated with heightened scrutiny. To be sure, the Court does not label its handiwork heightened scrutiny, and perhaps the method employed must hereafter be called “second order” rational-basis review rather than “heightened scrutiny.” But however labeled, the rational-basis test invoked today is most assuredly not the rational-basis test of Williamson v. Lee Optical of Oklahoma, Inc., 348 U. S. 483 (1955); Allied Stores of Ohio, Inc. v. Bowers, 358 U. S. 522 (1959), and their progeny.
The Court, for example, concludes that legitimate concerns for fire hazards or the serenity of the neighborhood do not justify singling out respondents to bear the burdens of these concerns, for analogous permitted uses appear to pose similar threats. Yet under the traditional and most minimal version of the rational-basis test, “reform may take one step at a time, addressing itself to the phase of the problem which seems most acute to the legislative mind.” Williamson v. Lee Optical of Oklahoma, Inc., supra, at 489; see American Federation of Labor v. American Sash Co., 335 U. S. 538 (1949); Semler v. Dental Examiners, 294 U. S. 608 (1935). The “record” is said not to support the ordinance’s classifications, ante, at 448, 450, but under the traditional standard we do not sift through the record to determine whether policy decisions are squarely supported by a firm factual foundation. Exxon Corp. v. Eagerton, 462 U. S. 176, 196 (1983); Minnesota v. Clover Leaf Creamery Co., 449 U. S. 456, 461-462, *459464 (1981); Firemen v. Chicago, R. I. & P. R. Co., 393 U. S. 129, 138-139 (1968). Finally, the Court further finds it “difficult to believe” that the retarded present different or special hazards inapplicable to other groups. In normal circumstances, the burden is not on the legislature to convince the Court that the lines it has drawn are sensible; legislation is presumptively constitutional, and a State “is not required to resort to close distinctions or to maintain a precise, scientific uniformity with reference” to its goals. Allied Stores of Ohio, Inc. v. Bowers, supra, at 527; see New Orleans v. Dukes, 427 U. S. 297, 303 (1976); Metropolis Theatre Co. v. City of Chicago, 228 U. S. 61, 68-70 (1913).
I share the Court’s criticisms of the overly broad lines that Cleburne’s zoning ordinance has drawn. But if the ordinance is to be invalidated for its imprecise classifications, it must be pursuant to more powerful scrutiny than the minimal rational-basis test used to review classifications affecting only economic and commercial matters. The same imprecision in a similar ordinance that required opticians but not optometrists to be licensed to practice, see Williamson v. Lee Optical of Oklahoma, Inc., supra, or that excluded new but not old businesses from parts of a community, see New Orleans v. Dukes, supra, would hardly be fatal to the statutory scheme.
The refusal to acknowledge that something more than minimum rationality review is at work here is, in my view, unfortunate in at least two respects. The suggestion that *460the traditional rational-basis test allows this sort of searching inquiry creates precedent for this Court and lower courts to subject economic and commercial classifications to similar and searching “ordinary” rational-basis review — a small and regrettable step back toward the days of Lochner v. New York, 198 U. S. 45 (1905). Moreover, by failing to articulate the factors that justify today’s “second order” rational-basis review, the Court provides no principled foundation for determining when more searching inquiry is to be invoked. Lower courts are thus left in the dark on this important question, and this Court remains unaccountable for its decisions employing, or refusing to employ, particularly searching scrutiny. Candor requires me to acknowledge the particular factors that justify invalidating Cleburne’s zoning ordinance under the careful scrutiny it today receives.
I — I HH
I have long believed the level of scrutiny employed m an equal protection case should vary with “the constitutional and societal importance of the interest adversely affected and the recognized invidiousness of the basis upon which the particular classification is drawn.” San Antonio Independent School District v. Rodriguez, 411 U. S. 1, 99 (1973) (Marshall, J., dissenting). See also Plyler v. Doe, 457 U. S. 202, 230-231 (1982) (Marshall, J., concurring); Dandridge v. Williams, 397 U. S. 471, 508 (1970) (Marshall, J., dissenting). When a zoning ordinance works to exclude the retarded from all residential districts in a community, these two considerations require that the ordinance be convincingly justified as substantially furthering legitimate and important purposes. Plyler, supra; Mississippi University for Women v. Hogan, 458 U. S. 718 (1982); Frontiero v. Richardson, 411 U. S. 677 (1973); Mills v. Habluetzel, 456 U. S. 91 (1982); see also Buchanan v. Warley, 245 U. S. 60 (1917).
*461First, the interest of the retarded in establishing group homes is substantial. The right to “establish a home” has long been cherished as one of the fundamental liberties embraced by the Due Process Clause. See Meyer v. Nebraska, 262 U. S. 390, 399 (1923). For retarded adults, this right means living together in group homes, for as deinstitutional-ization has progressed, group homes have become the primary means by which retarded adults can enter life in the community. The District Court found as a matter of fact that
“[t]he availability of such a home in communities is an essential ingredient of normal living patterns for persons who are mentally retarded, and each factor that makes such group homes harder to establish operates to exclude persons who are mentally retarded from the community.” App. to Pet. for Cert. A-8.
Excluding group homes deprives the retarded of much of what makes for human freedom and fulfillment — the ability to form bonds and take part in the life of a community.
Second, the mentally retarded have been subject to a “lengthy and tragic history,” University of California Regents v. Bakke, 438 U. S. 265, 303 (1978) (opinion of Powell, J.), of segregation and discrimination that can only be called grotesque. During much of the 19th century, mental retardation was viewed as neither curable nor dangerous and the retarded were largely left to their own devices. By the latter part of the century and during the first decades of the new one, however, social views of the retarded underwent a radical transformation. Fueled by the rising tide of Social Darwinism, the “science” of eugenics, and the extreme *462xenophobia of those years, leading medical authorities and others began to portray the “feebleminded” as a “menace to society and civilization . . . responsible in a large degree for many, if not all, of our social problems.” A regime of state-mandated segregation and degradation soon emerged that in its virulence and bigotry rivaled, and indeed paralleled, the worst excesses of Jim Crow. Massive custodial institutions were built to warehouse the retarded for life; the aim was to halt reproduction of the retarded and “nearly extinguish their race.” Retarded children were categorically excluded from *463public schools, based on the false stereotype that all were ineducable and on the purported need to protect nonretarded children from them. State laws deemed the retarded “unfit for citizenship.”
Segregation was accompanied by eugenic marriage and sterilization laws that extinguished for the retarded one of the “basic civil rights of man” — the right to marry and procreate. Skinner v. Oklahoma ex rel. Williamson, 316 U. S. 535, 541 (1942). Marriages of the retarded were made, and in some States continue to be, not only voidable but also often a criminal offense. The purpose of such limitations, which frequently applied only to women of child-bearing age, was unabashedly eugenic: to prevent the retarded from propagating. To assure this end, 29 States enacted compulsory eugenic sterilization laws between 1907 and 1931. J. Land-man, Human Sterilization 302-303 (1932). See Buck v. Bell, 274 U. S. 200, 207 (1927) (Holmes, J.); cf. Plessy v. Fergu*464son, 163 U. S. 537 (1896); Bradwell v. Illinois, 16 Wall. 130, 141 (1873) (Bradley, J., concurring in judgment).
Prejudice, once let loose, is not easily cabined. See University of California Regents v. Bakke, 438 U. S., at 395 (opinion of Marshall, J.). As of 1979, most States still categorically disqualified “idiots” from voting, without regard to individual capacity and with discretion to exclude left in the hands of low-level election officials. Not until Congress enacted the Education of the Handicapped Act, 84 Stat. 175, as amended, 20 U. S. C. §1400 et seq., were “the door[s] of public education” opened wide to handicapped children. Hendrick Hudson District Board of Education v. Rowley, 458 U. S. 176, 192 (1982). But most important, lengthy and continuing isolation of the retarded has perpetuated the ignorance, irrational fears, and stereotyping that long have plagued them.
In light of the importance of the interest at stake and the history of discrimination the retarded have suffered, the Equal Protection Clause requires us to do more than review the distinctions drawn by Cleburne’s zoning ordinance as if they appeared in a taxing statute or in economic or commercial legislation. The searching scrutiny I would give to re*465strictions on the ability of the retarded to establish community group homes leads me to conclude that Cleburne’s vague generalizations for classifying the “feeble-minded” with drug addicts, alcoholics, and the insane, and excluding them where the elderly, the ill, the boarder, and the transient are allowed, are not substantial or important enough to overcome the suspicion that the ordinance rests on impermissible assumptions or outmoded and perhaps invidious stereotypes. See Plyler v. Doe, 457 U. S. 202 (1982); Roberts v. United States Jaycees, 468 U. S. 609 (1984); Mississippi University for Women v. Hogan, 458 U. S. 718 (1982); Mills v. Habluetzel, 456 U. S. 91 (1982).
HH hH I — I
In its effort to show that Cleburne’s ordinance can be struck down under no “more exacting standard . . . than is normally accorded economic and social legislation,” ante, at 442, the Court offers several justifications as to why the retarded do not warrant heightened judicial solicitude. These justifications, however, find no support in our heightened-scrutiny precedents and cannot withstand logical analysis.
The Court downplays the lengthy “history of purposeful unequal treatment” of the retarded, see San Antonio Independent School District v. Rodriguez, 411 U. S., at 28, by pointing to recent legislative action that is said to “beli[e] a continuing antipathy or prejudice.” Ante, at 443. Building on this point, the Court similarly concludes that the retarded *466are not “politically powerless” and deserve no greater judicial protection than “[a]ny minority” that wins some political battles and loses others. Ante, at 445. The import of these conclusions, it seems, is that the only discrimination courts may remedy is the discrimination they alone are perspicacious enough to see. Once society begins to recognize certain practices as discriminatory, in part because previously stigmatized groups have mobilized politically to lift this stigma, the Court would refrain from approaching such practices with the added skepticism of heightened scrutiny.
Courts, however, do not sit or act in a social vacuum. Moral philosophers may debate whether certain inequalities are absolute wrongs, but history makes clear that constitutional principles of equality, like constitutional principles of liberty, property, and due process, evolve over time; what once was a “natural” and “self-evident” ordering later comes to be seen as an artificial and invidious constraint on human potential and freedom. Compare Plessy v. Ferguson, 163 U. S. 537 (1896), and Bradwell v. Illinois, supra, at 141 (Bradley, J., concurring in judgment), with Brown v. Board of Education, 347 U. S. 483 (1954), and Reed v. Reed, 404 U. S. 71 (1971). Shifting cultural, political, and social patterns at times come to make past practices appear inconsistent with fundamental principles upon which American society rests, an inconsistency legally cognizable under the Equal Protection Clause. It is natural that evolving standards of equality come to be embodied in legislation. When that occurs, courts should look to the fact of such change as a source of guidance on evolving principles of equality. In an analysis the Court today ignores, the Court reached this very conclusion when it extended heightened scrutiny to gender classifications and drew on parallel legislative developments to support that extension:
“[Ojver the past decade, Congress has itself manifested an increasing sensitivity to sex-based classifi*467cations [citing examples]. Thus, Congress itself has concluded that classifications based upon sex are inherently invidious, and this conclusion of a coequal branch of Government is not without significance to the question presently under consideration.” Frontiero v. Richardson, 411 U. S., at 687.
Moreover, even when judicial action has catalyzed legislative change, that change certainly does not eviscerate the underlying constitutional principle. The Court, for example, has never suggested that race-based classifications became any less suspect once extensive legislation had been enacted on the subject. See Patmore v. Sidoti, 466 U. S. 429 (1984).
For the retarded, just as for Negroes and women, much has changed in recent years, but much remains the same; outdated statutes are still on the books, and irrational fears or ignorance, traceable to the prolonged social and cultural isolation of the retarded, continue to stymie recognition of the dignity and individuality of retarded people. Heightened judicial scrutiny of action appearing to impose unnecessary barriers to the retarded is required in light of increasing recognition that such barriers are inconsistent with evolving principles of equality embedded in the Fourteenth Amendment.
The Court also offers a more general view of heightened scrutiny, a view focused primarily on when heightened scrutiny does not apply as opposed to when it does apply. Two *468principles appear central to the Court’s theory. First, heightened scrutiny is said to be inapplicable where individuals in a group have distinguishing characteristics that legislatures properly may take into account in some circumstances. Ante, at 441-442. Heightened scrutiny is also purportedly inappropriate when many legislative classifications affecting the group are likely to be valid. We must, so the Court says, “look to the likelihood that governmental action premised on a particular classification is valid as a general matter, not merely to the specifics of the case before us,” in deciding whether to apply heightened scrutiny. Ante, at 446.
If the Court’s first principle were sound, heightened scrutiny would have to await a day when people could be cut from a cookie mold. Women are hardly alike in all their characteristics, but heightened scrutiny applies to them because legislatures can rarely use gender itself as a proxy for these other characteristics. Permissible distinctions between persons must bear a reasonable relationship to their relevant characteristics, Zobel v. Williams, 457 U. S., at 70 (Brennan, J., concurring), and gender per se is almost never relevant. Similarly, that some retarded people have reduced capacities in some areas does not justify using retardation as a proxy for reduced capacity in areas where relevant individual variations in capacity do exist.
The Court’s second assertion — that the standard of review must be fixed with reference to the number of classifications to which a characteristic would validly be relevant — is similarly flawed. Certainly the assertion is not a logical one; that a characteristic may be relevant under some or even many circumstances does not suggest any reason to presume it relevant under other circumstances where there is reason to suspect it is not. A sign that says “men only” looks very *469different on a bathroom door than a courthouse door. But see Bradwell v. Illinois, 16 Wall. 130 (1873).
Our heightened-scrutiny precedents belie the claim that a characteristic must virtually always be irrelevant to warrant heightened scrutiny. Plyler, for example, held that the status of being an undocumented alien is not a “constitutional irrelevancy,” and therefore declined to review with strict scrutiny classifications affecting undocumented aliens. 457 U. S., at 219, n. 19. While Frontiero stated that gender “frequently” and “often” bears no relation to legitimate legislative aims, it did not deem gender an impermissible basis of state action in all circumstances. 411 U. S., at 686-687. Indeed, the Court has upheld some gender-based classifications. Rostker v. Goldberg, 453 U. S. 57 (1981); Michael M. v. Superior Court of Sonoma County, 450 U. S. 464 (1981). Heightened but not strict scrutiny is considered appropriate in areas such as gender, illegitimacy, or alienage because the Court views the trait as relevant under some circumstances but not others. That view — indeed the very concept of heightened, as opposed to strict, scrutiny — is flatly inconsistent with the notion that heightened scrutiny should not apply to the retarded because “mental retardation is a characteristic that the government may legitimately take into account in a wide range of decisions.” Ante, at 446. Because the government also may not take this characteristic into account in many circumstances, such as those presented here, careful review is required to separate the permissible from the invalid in classifications relying on retardation.
*470The fact that retardation may be deemed a constitutional irrelevancy in some circumstances is enough, given the history of discrimination the retarded have suffered, to require careful judicial review of classifications singling out the retarded for special burdens. Although the Court acknowledges that many instances of invidious discrimination against the retarded still exist, the Court boldly asserts that “in the vast majority of situations” special treatment of the retarded is “not only legitimate but also desirable.” Ante, at 444. That assertion suggests the Court would somehow have us calculate the percentage of “situations” in which a characteristic is validly and invalidly invoked before determining whether heightened scrutiny is appropriate. But heightened scrutiny has not been “triggered” in our past cases only after some undefined numerical threshold of invalid “situations” has been crossed. An inquiry into constitutional principle, not mathematics, determines whether heightened scrutiny is appropriate. Whenever evolving principles of equality, rooted in the Equal Protection Clause, require that certain classifications be viewed as potentially discriminatory, and when history reveals systemic unequal treatment, more searching judicial inquiry than minimum rationality becomes relevant.
Potentially discriminatory classifications exist only where some constitutional basis can be found for presuming that equal rights are required. Discrimination, in the Fourteenth Amendment sense, connotes a substantive constitutional judgment that two individuals or groups are entitled to be treated equally with respect to something. With regard to economic and commercial matters, no basis for such a conclusion exists, for as Justice Holmes urged the Lochner Court, the Fourteenth Amendment was not “intended to embody a particular economic theory . . . .” Lochner v. New York, 198 U. S., at 75 (dissenting). As a matter of substantive policy, therefore, government is free to move in any *471direction, or to change directions, in the economic and commercial sphere. The structure of economic and commercial life is a matter of political compromise, not constitutional principle, and no norm of equality requires that there be as many opticians as optometrists, see Williamson v. Lee Optical of Oklahoma, Inc., 348 U. S. 483 (1955), or-new businesses as old, see New Orleans v. Dukes, 427 U. S. 297 (1976).
But the Fourteenth Amendment does prohibit other results under virtually all circumstances, such as castes created by law along racial or ethnic lines, see Palmore v. Sidoti, 466 U. S., at 432-433; Loving v. Virginia, 388 U. S. 1 (1967); McLaughlin v. Florida, 379 U. S. 184 (1964); Shelley v. Kraemer, 334 U. S. 1, 23 (1948); Hernandez v. Texas, 347 U. S. 475 (1954), and significantly constrains the range of permissible government choices where gender or illegitimacy, for example, are concerned. Where such constraints, derived from the Fourteenth Amendment, are present, and where history teaches that they have systemically been ignored, a “more searching judicial inquiry” is required. United States v. Carotene Products Co., 304 U. S. 144, 153, n. 4 (1938).
That more searching inquiry, be it called heightened scrutiny or “second order” rational-basis review, is a method of *472approaching certain classifications skeptically, with judgment suspended until the facts are in and the evidence considered. The government must establish that the classification is substantially related to important and legitimate objectives, see, e. g., Craig v. Boren, 429 U. S. 190 (1976), so that valid and sufficiently weighty policies actually justify the departure from equality. Heightened scrutiny does not allow courts to second-guess reasoned legislative or professional judgments tailored to the unique needs of a group like the retarded, but it does seek to assure that the hostility or thoughtlessness with which there is reason to be concerned has not carried the day. By invoking heightened scrutiny, the Court recognizes, and compels lower courts to recognize, that a group may well be the target of the sort of prejudiced, thoughtless, or stereotyped action that offends principles of equality found in the Fourteenth Amendment. Where classifications based on a particular characteristic have done so in the past, and the threat that they may do so remains, heightened scrutiny is appropriate.
*473As the history of discrimination against the retarded and its continuing legacy amply attest, the mentally retarded have been, and in some areas may still be, the targets of action the Equal Protection Clause condemns. With respect to a liberty so valued as the right to establish a home in the community, and so likely to be denied on the basis of irrational fears and outright hostility, heightened scrutiny is surely appropriate.
IV
In light of the scrutiny that should be applied here, Cle-burne’s ordinance sweeps too broadly to dispel the suspicion that it rests on a bare desire to treat the retarded as outsiders, pariahs who do not belong in the community. The Court, while disclaiming that special scrutiny is necessary or warranted, reaches the same conclusion. Rather than striking the ordinance down, however, the Court invalidates it merely as applied to respondents. I must dissent from the novel proposition that “the preferred course of adjudication” *474is to leave standing a legislative Act resting on “irrational prejudice” ante, at 450, thereby forcing individuals in the group discriminated against to continue to run the Act’s gauntlet.
The Court appears to act out of a belief that the ordinance might be “rational” as applied to some subgroup of the retarded under some circumstances, such as those utterly without the capacity to live in a community, and that the ordinance should not be invalidated in toto if it is capable of ever being validly applied. But the issue is not “whether the city may never insist on a special use permit for the mentally retarded in an R-3 zone.” Ante, at 447. The issue is whether the city may require a permit pursuant to a blunderbuss ordinance drafted many years ago to exclude all the “feeble-minded,” or whether the city must enact a new ordinance carefully tailored to the exclusion of some well-defined subgroup of retarded people in circumstances in which exclusion might reasonably further legitimate city purposes.
By leaving the sweeping exclusion of the “feebleminded” to be applied to other groups of the retarded, the Court has created peculiar problems for the future. The Court does not define the relevant characteristics of respondents or their proposed home that make it unlawful to require them to seek a special permit. Nor does the Court delineate any principle that defines to which, if any, set of retarded people the ordinance might validly be applied. Cleburne’s City Council and retarded applicants are left without guidance as to the potentially valid, and invalid, applications of the ordinance. As a consequence, the Court’s as-applied remedy relegates future retarded applicants to the standardless discretion of low-level officials who have already shown an all too willing readiness to be captured by the “vague, undifferentiated fears,” ante, at 449, of ignorant or frightened residents.
Invalidating on its face the ordinance’s special treatment of the “feeble-minded,” in contrast, would place the responsibility for tailoring and updating Cleburne’s unconstitutional *475ordinance where it belongs: with the legislative arm of the city of Cleburne. If Cleburne perceives a legitimate need for requiring a certain well-defined subgroup of the retarded to obtain special permits before establishing group homes, Cleburne will, after studying the problem and making the appropriate policy decisions, enact a new, more narrowly tailored ordinance. That ordinance might well look very different from the current one; it might separate group homes (presently treated nowhere in the ordinance) from hospitals, and it might define a narrow subclass of the retarded for whom even group homes could legitimately be excluded. Special treatment of the retarded might be ended altogether. But whatever the contours such an ordinance might take, the city should not be allowed to keep its ordinance on the books intact and thereby shift to the courts the responsibility to confront the complex empirical and policy questions involved in updating statutes affecting the mentally retarded. A legislative solution would yield standards and provide the sort of certainty to retarded applicants and administrative officials that case-by-case judicial rulings cannot provide. Retarded applicants should not have to continue to attempt to surmount Cleburne’s vastly overbroad ordinance.
The Court’s as-applied approach might be more defensible under circumstances very different from those presented here. Were the ordinance capable of being cleanly severed, in one judicial cut, into its permissible and impermissible applications, the problems I have pointed out would be greatly reduced. Cf. United States v. Grace, 461 U. S. 171 (1983) (statute restricting speech and conduct in Supreme Court building and on its grounds invalid as applied to sidewalks); but cf. id., at 184-188 (opinion concurring in part and dissenting in part). But no readily apparent construction appears, nor has the Court offered one, to define which group of retarded people the city might validly require a permit of, and which it might not, in the R-3 zone. The Court’s as-applied holding is particularly inappropriate here, *476for nine-tenths of the group covered by the statute appears similarly situated to respondents, see ante, at 442, n. 9— a figure that makes the statutory presumption enormously overbroad. Cf. Stanley v. Illinois, 405 U. S. 645 (1972) (invalidating statutory presumption despite State’s insistence that it validly applied to “most” of those covered).
To my knowledge, the Court has never before treated an equal protection challenge to a statute on an as-applied basis. When statutes rest on impermissibly overbroad generalizations, our cases have invalidated the presumption on its face. We do not instead leave to the courts the task of redrafting the statute through an ongoing and cumbersome process of “as applied” constitutional rulings. In Cleveland Board of Education v. LaFleur, 414 U. S. 632 (1974), for *477example, we invalidated, inter alia, a maternity leave policy that required pregnant schoolteachers to take unpaid leave beginning five months before their expected due date. The school board argued that some teachers became physically incapable of performing adequately in the latter stages of their pregnancy, and we accepted this justification for purposes of our decision. Assuming the policy might validly be applied to some teachers, particularly in the last few weeks of their pregnancy, id., at 647, n. 13, we nonetheless invalidated it in toto, rather than simply as applied to the particular plaintiff. The Court required school boards to employ “alternative administrative means” to achieve their legitimate health and safety goal, id., at 647, or the legislature to enact a more carefully tailored statute, id., at 647, n. 13.
Similarly, Caban v. Mohammed, 441 U. S. 380 (1979), invalidated a law that required parental consent to adoption from unwed mothers but not from unwed fathers. This distinction was defended on the ground, inter alia, that unwed fathers were often more difficult to locate, particularly during a child’s infancy. We suggested the legislature might make proof of abandonment easier or proof of paternity harder, but we required the legislature to draft a new statute tailored more precisely to the problem of locating unwed fathers. The statute was not left on the books by invalidating it only as applied to unwed fathers who actually proved they could be located. When a presumption is unconstitutionally overbroad, the preferred course of adjudication is to strike it down. See also United States Dept. of Agriculture v. Moreno, 413 U. S. 528 (1973); Stanley v. Illinois, supra; Vlandis v. Kline, 412 U. S. 441, 453-454 (1973); Carrington v. Rash, 380 U. S. 89 (1965); Sugarman v. Dougall, 413 U. S. 634, 646-649 (1973); Weber v. Aetna Casualty & Surety Co., 406 U. S. 164 (1972); Levy v. Louisiana, 391 U. S. 68 (1968).
In my view, the Court’s remedial approach is both unprecedented in the equal protection area and unwise. This doc*478trinal change of course was not sought by the parties, suggested by the various amici, or discussed at oral argument. Moreover, the Court does not persuasively reason its way to its novel remedial holding nor reconsider our prior cases directly on point. Instead, the Court simply asserts that “this is the preferred course of adjudication.” Given that this assertion emerges only from today’s decision, one can only hope it will not become entrenched in the law without fuller consideration.
V
The Court’s opinion approaches the task of principled equal protection adjudication in what I view as precisely the wrong way. The formal label under which an equal protection claim is reviewed is less important than careful identification of the interest at stake and the extent to which society recognizes the classification as an invidious one. Yet in focusing obsessively on the appropriate label to give its standard of review, the Court fails to identify the interests at stake or to articulate the principle that classifications based on mental retardation must be carefully examined to assure they do not rest on impermissible assumptions or false stereotypes regarding individual ability and need. No guidance is thereby given as to when the Court’s freewheeling, and potentially dangerous, “rational-basis standard” is to be employed, nor is attention directed to the invidiousness of grouping all retarded individuals together. Moreover, the Court’s narrow, as-applied remedy fails to deal adequately with the overbroad presumption that lies at the heart of this case. Rather than leaving future retarded individuals to run the gauntlet of this overbroad presumption, I would affirm the judgment of the Court of Appeals in its entirety and would strike down on its face the provision at issue. I therefore concur in the judgment in part and dissent in part.