dissenting.
The Court’s disposition of this case is objectionable on two grounds. First is its conclusion that men challenging a gender-based statute which treats them less favorably than women may invoke a more stringent standard of judicial review than pertains to most other types of classifications. Second is the Court’s enunciation of this standard, without citation to any source, as being that “classifications by gender must serve important governmental objectives and must be substantially related to achievement of those objectives.” Ante, at 197 (emphasis added). . The only redeeming feature of the Court’s opinion, to my mind, is that it apparently signals a retreat by those who joined the plurality opinion in Frontiero v. Richardson, 411 U. S. 677 (1973), from their view that sex is a “suspect” classification for purposes of equal protection analysis. I think the Oklahoma statute challenged here need pass only the “rational basis” equal *218protection analysis expounded in cases such as McGowan v. Maryland, 366 U. S. 420 (1961), and Williamson v. Lee Optical Co., 348 U. S. 483 (1955), and I believe that it is constitutional under that analysis.
I
In Frontiero v. Richardson, supra, the opinion for the plurality sets forth the reasons of four Justices for concluding that sex should be regarded as a suspect classification for purposes of equal protection analysis. These reasons center on our Nation’s “long and unfortunate history of sex discrimination,” 411 U. S., at 684, which has been reflected in a whole range of restrictions on the legal rights of women, not the least of which have concerned the ownership of property and participation in the electoral process. Noting that the pervasive and persistent nature of the discrimination experienced by women is in part the result of their ready identifiability, the plurality rested its invocation of strict scrutiny largely upon the fact that “statutory distinctions between the sexes often have the effect of invidiously relegating the entire class of females to inferior legal status without regard to the actual capabilities of its individual members.” Id., at 686-687. See Stanton v. Stanton, 421 U. S. 7, 14-15 (1975).
Subsequent to Frontiero, the Court has declined to hold that sex is a suspect class, Stanton v. Stanton, supra, at 13, and no such holding is imported by the Court’s resolution of this case. However, the Court’s application here of an elevated or “intermediate” level scrutiny, like that invoked in cases dealing with discrimination against females, raises the question of why the statute here should be treated any differently from countless legislative classifications unrelated to sex which have been upheld under a minimum rationality standard. Jefferson v. Hackney, 406 U. S. 535, 546-547 (1972); Richardson v. Belcher, 404 U. S. 78, 81-84 (1971); Dandridge v. Williams, 397 U. S. 471, 484-485 (1970); *219McGowan v. Maryland, supra, at 425-426; Flemming v. Nestor, 363 U. S. 603, 611 (1960); Williamson v. Lee Optical Co., supra, at 488-489.
Most obviously unavailable to support any kind of special scrutiny in this case, is a history or pattern of past discrimination, such as was relied on by the plurality in Frontiero to support its invocation of strict scrutiny. There is no suggestion in the Court’s opinion that males in this age group are in any way peculiarly disadvantaged, subject to systematic discriminatory treatment, or otherwise in need of special solicitude from the courts.
The Court does not discuss the nature of the right involved, and there is no reason to believe that it sees the purchase of 3.2% beer as implicating any important interest, let alone one that is “fundamental” in the constitutional sense of invoking strict scrutiny. Indeed, the Court’s accurate observation that the statute affects the selling but not the drinking of 3.2% beer, ante, at 204, further emphasizes the limited effect that it has on even those persons in the age group involved. There is, in sum, nothing about the statutory classification involved here to suggest that it affects an interest, or works against a group, which can claim under the Equal Protection Clause that it is entitled to special judicial protection.
It is true that a number of our opinions contain broadly phrased dicta implying that the same test should be applied to all classifications based on sex, whether affecting females or males. E. g., Frontiero v. Richardson, supra, at 688; Reed v. Reed, 404 U. S. 71, 76 (1971). However, before today, no decision of this Court has applied an elevated level of scrutiny to invalidate a statutory discrimination harmful to males, except where the statute impaired an important personal interest protected by the Constitution.1 There being no such interest *220here, and there being no plausible argument that this is a discrimination against .females,2 the Court’s reliance on our previous sex-discrimination cases is ill-founded. It treats gender classification as a talisman which — without regard to the rights involved or the persons affected- — calls into effect a heavier burden of judicial review.
The Court’s conclusion that a law which treats males less favorably than females "must serve important governmental objectives and must be substantially related to achievement of those objectives” apparently comes out of thin air. The Equal Protection Clause contains no such language, and none of our previous cases adopt that standard. I would think we have had enough difficulty with the two standards of review which our cases have recognized — the *221norm of “rational basis,” and the “compelling state interest” required where a “suspect classification” is involved — so as to counsel weightily against the insertion of still another “standard” between those two. How is this Court to divine what objectives are important? How is it to determine whether a particular law is “substantially” related to the achievement of such objective, rather than related in some other way to its achievement? Both of the phrases used are so diaphanous and elastic as to invite subjective judicial preferences or prejudices relating to particular types of legislation, masquerading as judgments whether such legislation is directed at “important” objectives or, whether the relationship to those objectives is “substantial” enough.
I would have thought that if this Court were to leave anything to decision by the popularly elected branches of the Government, where no constitutional claim other than that of equal protection is invoked, it would be the decision as to what governmental objectives to be achieved by law are “important,” and which are not. As for the second part of the Court’s new test, the Judicial Branch is probably in no worse position than the Legislative or Executive Branches to determine if there is any rational relationship between a classification and the purpose which it might be thought to serve. But the introduction of the adverb “substantially” requires courts to make subjective judgments as to operational effects, for which neither their expertise nor their access to data fits them. And even if we manage to avoid both confusion and the mirroring of our own preferences in the development of this new doctrine, the thousands of judges in other courts who must interpret the Equal Protection Clause may not be so fortunate.
II
The applicable ra.tional-basis test is one which
“permits the States a wide scope of discretion in enacting laws which affect some groups of citizens differently than *222others. The constitutional safeguard is offended only if the classification rests on grounds wholly irrelevant to the achievement of the State's objective. State legislatures are presumed to have acted within their constitutional power despite the fact that, in practice, their laws result in some inequality. A statutory discrimination will not be set aside if any state of facts reasonably may be conceived to justify it.” McGowan v. Maryland, 366 U. S., at 425-426 (citations omitted).
Our decisions indicate that application of the Equal Protection Clause in a context not justifying an elevated level of scrutiny does not demand “mathematical nicety” or the elimination of all inequality. Those cases recognize that the practical problems of government may require rough accommodations of interests, and hold that such accommodations should be respected unless no reasonable basis can be found to support them. Dandridge v. Williams, 397 U. S., at 485. Whether the same ends might have been better or more precisely served by a different approach is no part of the judicial inquiry under the traditional minimum rationality approach. Richardson v. Belcher, 404 U. S., at 84.
The Court “accept[s] for purposes of discussion” the District Court's finding that the purpose of the provisions in question was traffic safety, and proceeds to examine the statistical evidence in the record in order to decide if “the gender-based distinction closely serves to achieve that objective.” Ante, at 199, 200 (emphasis added).. (Whether there is a difference between laws which “closely serv[e]” objectives and those which are only “substantially related” to their achievement, ante, at 197, we are not told.) I believe that a more traditional type of scrutiny is appropriate in this case, and I think that the Court would have done well here to heed its own warning that “[i]t is unrealistic to expect .. . members of the judiciary ... to be well versed in the rigors of experimental or statistical technique.” Ante, at 204. One *223need not immerse oneself in the fine points of statistical analysis, however, in order to see the weaknesses in the Court’s attempted denigration of the evidence at hand.
One survey of arrest statistics assembled in 1973 indicated that males in the 18-20 age group were arrested for “driving under the influence” almost 18 times as often as their female counterparts, and for “drunkenness” in a ratio of almost 10 to l.3 Accepting, as the Court does, appellants’ comparison of the total figures with 1973 Oklahoma census data, this survey indicates a 2% arrest rate among males in the age group, as compared to a .18% rate among females.
Other surveys indicated (1) that over the five-year period from 1967 to 1972, nationwide arrests among those under 18 for drunken driving increased 138%, and that 93% of all persons arrested for drunken driving were male; 4 (2) that youths in the 17-21 age group were overrepresented among those killed or injured in Oklahoma traffic accidents, that male casualties substantially exceeded female, and that deaths in this age group continued to rise while overall traffic deaths declined;5 (3) that over three-fourths of the drivers under 20 in the Oklahoma City area are males, and that each of them, on average, drives half again as many miles per year as their female counterparts;6 (4) that four-fifths of male drivers *224under 20 in the Oklahoma City area state a drink preference for beer, while about three-fifths of female drivers of that age state the same preference;7 and (5) that the percentage of male drivers under 20 admitting to drinking within two hours of driving was half again larger than the percentage for females, and that the percentage of male drivers of that age group with a blood alcohol content greater than .01% was almost half again larger than for female drivers.8
The Court's criticism of the statistics relied on by the District Court conveys the impression that a legislature in enacting a new law is to be subjected to the judicial equivalent of a doctoral examination in statistics. Legislatures are not held to any rules of evidence such as those which may govern courts or other administrative bodies, and are entitled to draw factual conclusions on the basis of the determination of probable cause which an arrest by a police officer normally represents. In this situation, they could reasonably infer that the incidence of drunk driving is a good deal higher than the incidence of arrest.
And while, as the Court observes, relying on a report to a Presidential Commission which it cites in a footnote, such statistics may be distorted as a result of stereotyping, the legislature is not required to prove before a court that its statistics are perfect. In any event, if stereotypes are as pervasive as the Court suggests, they may in turn influence the conduct of the men and women in question, and cause the young men to conform to the wild and reckless image which is their stereotype.
The Court also complains of insufficient integration of the various surveys on several counts — that the injury and death figures are in no way directly correlated with intoxication, ante, at 201 n. 9; that the national figures for drunk driving contain no breakdown for the 18-21-year-old group, *225ante, at 201 n. 11; and that the arrest records for intoxication are not tied to the consumption of 3.2% beer, ante, at 201-202, nn. 11 and 12. But the State of Oklahoma — and certainly this Court for purposes of equal protection review — can surely take notice of the fact that drunkenness is a significant cause of traffic casualties, and that youthful offenders have participated in the increase of the drunk-driving problem. On this latter point, the survey data indicating increased driving casualties among 18-21-year-olds, while overall casualties dropped, are not irrelevant.
Nor is it unreasonable to conclude from the expressed preference for beer by four-fifths of the age-group males that that beverage was a predominant source of their intoxication-related arrests. Taking that as the predicate, the State could reasonably bar those males from any purchases of alcoholic beer, including that of the 3.2% variety. This Court lacks the expertise or the data to evaluate the intoxicating properties of that beverage, and in that posture our only appropriate course is to defer to the reasonable inference supporting the statute — that taken in sufficient quantity this beer has the same effect as any alcoholic beverage.
Quite apart from these alleged methodological deficiencies in the statistical evidence, the Court appears to hold that that evidence, on its face, fails to support the distinction drawn in the statute. The Court notes that only 2% of males (as against .18% of females) in the age group were arrested for drunk driving, and that this very low figure establishes “an unduly tenuous ‘fit’ ” between maleness and drunk driving in the 18-20-year-old group. On this point the Court misconceives the nature of the equal protection inquiry.
The rationality of a statutory classification for equal protection purposes does not depend upon the statistical “fit” between the class and the trait sought to be singled out. It turns on whether there may be a sufficiently higher in*226cidence of the trait within the included class than in the excluded class to justify different treatment. Therefore the present equal protection challenge to this gender-based discrimination poses only the question whether the incidence of drunk driving among young men is sufficiently greater than among young women to justify differential treatment. Notwithstanding the Court’s critique of the statistical evidence, that evidence suggests clear differences between the drinking and driving habits of young men and women. Those differences are grounds enough for the State reasonably to conclude that young males pose by far the greater drunk-driving hazard, both in terms of sheer numbers and in terms of hazard on a per-driver basis. The gender-based difference in treatment in this case is therefore not irrational.
The Court’s argument that a 2% correlation between maleness and drunk driving is constitutionally insufficient therefore does not pose an equal protection issue concerning discrimination between males and females. The clearest demonstration of this is the fact that the precise argument made by the Court would be equally applicable to a flat bar on such purchases by anyone, male or female, in the 18-20 age group; in fact it would apply a fortiori in that case given the even more “tenuous 'fit’ ” between drunk-driving arrests and femaleness. The statistics indicate that about 1 % of the age group population as a whole is arrested. What the Court’s argument is relevant to is not equal protection, but due process — whether there are enough persons in the category who drive while drunk to justify a bar against purchases by all members of the group.
Cast in those terms, the argument carries little weight, in light of our decisions indicating that such questions call for a balance of the State’s interest against the harm resulting from any overinclusiveness or underinclusiveness. Vlandis v. Kline, 412 U. S. 441, 448-452 (1973). The personal interest harmed *227here is very minor — the present legislation implicates only the right to purchase 3.2% beer, certainly a far cry from the important personal interests which have on occasion supported this Court’s invalidation of statutes on similar reasoning. Cleveland Board of Education v. LaFleur, 414 U. S. 632, 640 (1974); Stanley v. Illinois, 405 U. S. 645, 651 (1972). And the state interest involved is significant — the prevention of injury and death on the highways.
This is not a case where the classification can only be justified on grounds of administrative convenience. Vlandis v. Kline, supra, at 451; Stanley v. Illinois, supra, at 656. There being no apparent way to single out persons likely to drink and drive, it seems plain that the legislature was faced here with the not atypical legislative problem of legislating in terms of broad categories with regard to the purchase and consumption of alcohol. I trust, especially in light of the Twenty-first Amendment, that there would be no due process violation if no one in this age group were allowed to purchase 3.2% beer. Since males drink and drive at a higher rate than the age group as a whole, I fail to see how a statutory bar with regard only to them can create any due process problem.
The Oklahoma Legislature could have believed that 18-20-year-old males drive substantially more, and tend more often to be intoxicated than their female counterparts; that they prefer beer and admit to drinking and driving at a higher rate than females; and that they suffer traffic injuries out of proportion to the part they make up of the population. Under the appropriate rational-basis test for equal protection, it is neither irrational nor arbitrary to bar them from making purchases of 3.2% beer, which purchases might in many cases be made by a young man who immediately returns to his vehicle with the beverage in his possession. The record does not give any good indication of the true proportion of males in the age group who drink and drive (ex*228cept that it is no doubt greater than the 2% who are arrested), but whatever it may be I cannot see that the mere purchase right involved could conceivably raise a due process question. There being no violation of either equal protection or due process, the statute should accordingly be upheld.
In Stanley v. Illinois, 405 U. S. 645 (1972), the Court struck down a statute allowing separation of illegitimate children from a surviving father *220but not a surviving mother, without any showing of parental unfitness. The Court stated that “the interest of a parent in the companionship, care, custody, and management of his or her children ‘eome[s] to this Court with a momentum for respect lacking when appeal is made to liberties which derive merely from shifting economic arrangements.’ ”
In Kahn v. Shevin, 416 U. S. 351 (1974), the Court upheld Florida’s $500 property tax exemption for widows only. The opinion of the Court appears to apply a rational-basis test, id., at 355, and is so understood by the dissenters. Id., at 357 (Brennan, J., joined by Marshall, J., dissenting).
In Weinberger v. Wiesenfeld, 420 U. S. 636 (1975), the Court invalidated § 202 (g) of the Social Security Act, which allowed benefits to mothers but not fathers of minor children, who survive the wage earner. This statute was treated, in the opinion of the Court, as a discrimination against female wage earners, on the ground that it minimizes the financial security which their work efforts provide for their families. 420 U. S., at 645.
I am not unaware of the argument from time to time advanced, that all discriminations between the sexes ultimately redound to the detriment of females, because they tend to reinforce “old notions” restricting the roles and opportunities of women. As a general proposition applying equally to all sex categorizations, I believe that this argument was implicitly found to carry little weight in our decisions upholding gender-based differences. See Schlesinger v. Ballard, 419 U. S. 498 (1975); Kahn v. Shevin, supra. Seeing no assertion that it has special applicability to the situation at hand, I believe it can be dismissed as an insubstantial consideration.
Extract from: Oklahoma Bureau of Investigation, Arrest Statistics for September, October, November, and December 1973. Dgfendants’ Exhibit 1, Jurisdictional Statement A22. Extract from: Oklahoma City Police Department, Arrest Statistics for 1973. Defendants’ Exhibit 2, Jurisdictional Statement A23. See ante, at 200 n. 8.
Extract from: Federal Bureau of Investigation, Crime in the United States, 1972. Defendants’ Exhibit 6, App. 182-184.
Extract from: Oklahoma Department of Public Safety, Summary of Statewide Collisions for 1972, 1973. Defendants’ Exhibits 4 and 5, Jurisdictional Statement A30-A31.
Extract from: Oklahoma Management and Engineering Consulting, Inc., Report to Alcohol Safety Action Program (1973). Defendants’ Exhibit 3, Table 1, Jurisdictional Statement A25.
Id., at A27 (Table 3), A29 (Table 5).
Id., at A25 (Table 1). See ante, at 203 n. 16.