dissenting.
Although the Court purports to apply settled principles to unique facts, the result reached does not square with either Title YII or the Equal Protection Clause. Accordingly, but respectfully, I dissent.
I
As an initial matter, the Court is unwise in failing to remand the statutory claims to the Court of Appeals. The District Court decided the Title VII issue only because it provided a basis for allowing attorney’s fees. 414 F. Supp. 277, 278 (SDNY 1976). The Court of Appeals did not deal with Title VII, relying instead on the intervening passage of the Civil Rights Attorney’s Fees Awards Act of 1976,1 which authorized the award of fees for success on the equal protection claim today held infirm by the Court. 558 F. 2d 97, 99-100 (CA2 1977). In such circumstances, on finding that we disagree with the judgment of the Court of Appeals as to the constitutional question, we would usually remand the unexplored alternative basis for relief.2 E. g., Vermont Yankee *598Nuclear Power Corp. v. NRDC, 435 U. S. 519, 549 (1978). And see Arlington Heights v. Metropolitan Housing Dev. Corp., 429 U. S. 252, 271 (1977), which involved nearly identical circumstances. That course would obviate the need for us to deal with what the Court considers to be a factual issue or at least would provide assistance in analyzing the issue.
Because the Court has decided the question, however, I must express my reservations about the merits of that decision. In a disparate-impact hiring case such as this, the plaintiff must show that the challenged practice excludes members of a protected group in numbers disproportionate to their incidence in the pool of potential employees.3 Respondents made out a sufficient, though not strong, prima facie case by proving that about 63% of those using methadone in the New York City area are black or Hispanic and that only about 20% of the relevant population, as a whole belongs to one of those groups.4 ■ I think it fair to conclude, as the District Court must *599have, that blacks and Hispanics suffer three times as much from the operation of the challenged rule excluding methadone users as one would expect from a neutral practice. Thus, excluding those who are or have been in methadone programs “operate [s] to render ineligible a markedly disproportionate number” of blacks and Hispanics. Griggs v. Duke Power Co., 401 U. S. 424, 429 (1971).
In response to this, the Court says that the 63% statistic was not limited to those who worked for or sought to work for petitioners and to those who have been successfully maintained on methadone, and that it does not include those in private clinics. Ante, at 584-586. I suggest, in the first place, that these attacks on facially valid statistics should have been made in the District Court and the Court of Appeals, see Dothard v. Rawlinson, 433 U. S. 321, 331 (1977); the first contention was not even made in this Court. It also seems to me that petitioners have little to complain about insofar as the makeup of the applicant pool is concerned since they refused on grounds of irrelevancy to allow discovery of the racial background of the applicants denied employment pursuant to the methadone rule.
In any event, I cannot agree with the Court’s assertions that this evidence “reveals little if anything,” “tells us nothing,” and is “virtually irrelevant.” Ante, at 585-586.5 There is not a *600shadow of doubt that methadone users do apply for employment with petitioners, and because 63% of all methadone users are black or Hispanic, there is every reason to conclude that a majority of methadone users who apply are also from these minority groups. Almost 5% of all applicants are rejected due to the rule, and undoubtedly many black and Hispanic methadone users are among those rejected. Why would proportionally fewer of them than whites secure work with petitioners absent the challenged practice? The Court gives no reason whatsoever for rejecting this sensible inference, and where the inference depends so much on local knowledge, I would accept the judgment of the District Court rather than purport to make an independent judgment from the banks of the Potomac. At the very least, as I have said, I would seek the views of the Court of Appeals.
The Court complains that even if minority groups make up 63% of methadone-user applicants this statistic is an insufficient indicator of the composition of the group found by the District Court to have been wrongly excluded — that is, those who have been successfully maintained for a year or more. I cannot, however, presume with the Court that blacks or Hispanics will be less likely than whites to succeed on methadone. I would have thought the presumption, until rebutted, would be one of an equal chance of success, and there has been no rebuttal.
Finally, as to the racial composition of the patients at private clinics, I note first that the District Court found that “[bjetween 62% and 65% of methadone maintained persons in New York City are black and Hispanic . . . 414 F. Supp., at 279/ The finding was for the total population, not just for public clinics. Even assuming that the Court wishes to overturn this finding of fact as clearly erroneous, I see no support for doing so. The evidence from the Methadone Information Center at Rockefeller University indicated that 61% of all patients in the metropolitan area were black or Puerto Rican (with 5.85% undefined). This was based on a *6011,400-patient sample, which, according to the Center, “was drawn on a random basis and very accurately reflects the total population for Metropolitan New York City” (emphasis supplied). There is no reason to believe that this study, which in its reporting of the total number of patients of all races included both public and private clinics, did not include private programs in its racial-composition figures.6 And even if everyone in the private clinics were white, a highly unlikely assumption at best,7 the challenged rule would still automatically exclude a substantially greater number of blacks and Hispanics than would a practice with a racially neutral effect.
With all due respect, I would accept the statistics as making *602a prima facie case of disparate impact. Obviously, the case could have been stronger, but this Court is unjustified in displacing the District Court’s acceptance of uncontradicted, relevant evidence. Perhaps sensing that, the Court goes on to say that if such a prima facie showing was made it was rebutted by the fact that the rule is “job related.”
Petitioners had the burden of showing job relatedness. They did not show that the rule results in a higher quality labor force, that such a labor force is necessary, or that the cost of making individual decisions about those on methadone was prohibitive. Indeed, as shown in the equal protection discussion infra, petitioners have not come close to showing that the present rule is “demonstrably a reasonable measure of job performance.” Griggs, 401 U. S., at 436. No one could reasonably argue that petitioners have made the kind of showing demanded by Griggs or Albemarle Paper Co. v. Moody, 422 U. S. 405 (1975). By petitioners’ own stipulation, see n. 14, infra, this employment barrier was adopted “without meaningful study of [its] relationship to job-performance ability.” Griggs, supra, at 431. As we stated in Washington v. Davis, 426 U. S. 229, 247 (1976), Title VII “involves a more probing judicial review of, and less deference to, the seemingly reasonable acts of administrators and executives than is appropriate under the Constitution . . . .” Therefore, unlike the majority, ante, at 587 n. 31, I think it insufficient that the rule as a whole has some relationship to employment so long as a readily identifiable and severable part of it does not.
II
I also disagree with the Court’s disposition of the equal protection claim in light of the facts established below. The District Court found that the evidence conclusively established that petitioners exclude from employment all persons who are successfully on methadone maintenance — that is, those who after one year are “free of the use of heroin, other illicit *603drugs, and problem drinking,” 399 F. Supp. 1032, 1047 (SDNY 1975) — and those who have graduated from methadone programs and remain drug free for less than five years;8 that past *604or present successful methadone maintenance is not a meaningful predictor of poor performance or conduct in most job categories; that petitioners could use their normal employee-screening mechanisms to separate the successfully maintained users from the unsuccessful; and that petitioners do exactly that for other groups that common sense indicates might also be suspect employees.9 Petitioners did not challenge these factual conclusions in the Court of Appeals, but that court nonetheless reviewed the evidence and found that it overwhelmingly supported the District Court’s findings. 558 F. 2d, at 99. It bears repeating, then, that both the District Court and the Court of Appeals found that those who have been maintained on methadone for at least a year and who are free from the use of illicit drugs and alcohol can easily be identified through normal personnel procedures and, for a great many jobs, are as employable as and present no more risk than applicants from the general population.
Though petitioners’ argument here is primarily an attack upon the factfinding below, the Court does not directly accept that thesis. Instead, it concludes that the District Court and the Court of Appeals both misapplied the Equal Protection *605Clause. On the facts as found, however, one can reach the Court’s result only if that Clause imposes, no real constraint at all in this situation.
The question before us is the rationality of placing successfully maintained or recently cured persons in the same category as those just attempting to escape heroin addiction or who have failed to escape it, rather than in with the general population.10 The asserted justification for the challenged classification is the objective of a capable and reliable work force, and thus the characteristic in question is employability. “Employability,” in this regard, does not mean that any particular applicant, much less every member of a given group of applicants, will turn out to be a model worker. Nor does it mean that no such applicant will ever become or be discovered to be a malingerer, thief, alcoholic, or even heroin addict. All employers take such risks. Employability, as the District Court used it in reference to successfully maintained methadone users, means only that the employer is no more likely to find a member of that group to be an unsatisfactory employee than he would an employee chosen from the general population.
Petitioners had every opportunity, but presented nothing to negative the employability of successfully maintained methadone users as distinguished from those who were unsuccessful. Instead, petitioners, like the Court, dwell on the methadone failures — those who quit the programs or who remain but turn to illicit drug use. The Court, for instance, makes much of the drug use of many of those in methadone programs, including those who have been in such programs for more than one year. Ante, at 576, and n. 10. But this has little force *606since those persons are not “successful,” can be and have been identified as such, see ante, at 574-575,11 and, despite the Court’s efforts to put them there, see ante, at 590 n. 33, are not within the protection of the District Court’s injunction. That 20% to 30% are unsuccessful after one year in a methadone program tells us nothing about the employability of the successful group, and it is the latter category of applicants that the District Court and the Court of Appeals held to be unconstitutionally burdened by the blanket rule disqualifying them from employment.
The District Court and the Court of Appeals were therefore fully justified in finding that petitioners could not reasonably have concluded that the protected group is less employable than the general population and that excluding it “has no rational relation to the demands of the jobs to be performed.” 12 399 F. Supp., at 1057. In fact, the Court assumes that petitioners’ policy is unnecessarily broad in excluding the successfully maintained and the recently cured, ante, at 592, and that a member of that group can be selected with adequate precision. Ante, at 574D575. Despite this, the validity of the exclusion is upheld on the rational basis of the uninvolved portion of the rule, that is, that the rule excludes many who are less employable. But petitioners must justify the distinction between groups, not just the policy to which they have attached the classification. The purpose of the rule as a whole is *607relevant only if the classification within the rule serves the purpose, but the majority’s assumption admits that is not so.
Justification of the blanket exclusion is not furthered by the statement that “any special rule short of total exclusion ... is likely to be less precise” than the current rule. Ante, at 590. If the rule were narrowed as the District Court ordered, it would operate more precisely in at least one respect, for many employable persons would no longer be excluded. Nor does the current rule provide a “bright line,” for there is nothing magic about the point five years after treatment has ended. There is a risk of “regression” among those who have never used methadone, and the Court cannot overcome the District Court’s finding that a readily ascertainable point exists at which the risk has so decreased that the maintained or recently cured person is generally as employable as anyone else.13
Of course, the District Court’s order permitting total exclusion of all methadone users maintained for less than one year, whether successfully or not, would still exclude some employables and would to this extent be overinclusive. “Over-inclusiveness” as to the primary objective of employability is accepted for less successful methadone users because it fulfills a secondary purpose and thus is not “overinclusive” at all. See Vance v. Bradley, ante, at 109. Although many of those who have not been successfully maintained for a year are employable, as a class they, unlike the protected group, are not as employable as the general population. Thus, even assuming the bad risks could be identified, serving the end of employability would require unusual efforts to determine those more likely to revert. But that legitimate *608secondary goal is not fulfilled by excluding the protected class: The District Court found that the fact of successful participation for one year could be discovered through petitioners’ normal screening process without additional effort and, I repeat, that those who meet that criterion are no more likely than the average applicant to turn out to be poor employees.14 Ac*609cordingly, the rule’s classification of successfully maintained persons as dispositively different from the general population is left without any justification and, with its irrationality and invidiousness thus uncovered, must fall before the Equal Protection Clause.15
*610Finally, even were the District Court wrong, and even were successfully maintained persons marginally less employable than the average applicant,16 the blanket exclusion of only these people, when but a few are actually unemployable and when many other groups have varying numbers of unemployable members, is arbitrary and unconstitutional. Many per*611sons now suffer from or may again suffer from some handicap related to employability.17 But petitioners have singled out respondents — unlike ex-offenders, former alcoholics and mental patients, diabetics, epileptics, and those currently using tranquilizers, for example — for sacrifice to this at best ethereal and likely nonexistent risk of increased unemployability. Such an arbitrary assignment of burdens among classes that are similarly situated with respect to the proffered objectives is the type of invidious choice forbidden by the Equal Protection Clause.18
U. S. C. § 1988.
The Court finds it inappropriate to remand because the Title YII question “was fully aired before the District Court, . . . involves the application of settled legal principles to uncontroversial facts, and . . . *598has been carefully briefed in this Court without any of the parties’ even suggesting the possibility of a remand.” Ante, at 583-584, n. 24. The Court is able to overturn the Title YII judgment below, however, only after reversing some of the District Court’s key findings of fact, which the parties strongly contest, on grounds that were not aired at all in the District Court or the Court of Appeals. See n. 4, infra, and infra, at 600 and n. 6.
See ante, at 584; Dothard v. Rawlinson, 433 U. S. 321, 329 (1977). The failure to hire is not “because of” race, color, religion, sex, or national origin if the adverse relationship of the challenged practice to one of those factors is purely a matter of chance — a statistical coincidence. See Griggs v. Duke Power Co., 401 U. S. 424, 430 (1971); Civil Rights Act of 1964, §703 (a), 42 U. S. C. §2000e-2 (a). Beyond the statistically significant relationship between race and participation in methadone programs shown by the figures here, respondents introduced direct evidence that the high frequency of minorities among the disqualified group was not just a chance aberration. See nn. 7 and 15, infra.
The Court asserts that the proper percentage is 36.3. Respondents relied upon the 1970 census figures for the New York Standard Metropolitan Statistical Area work force: 15.0% black and 5.1% Hispanic. Petitioners accept the 20% figure. Brief for Petitioners 53. And the *599District Court apparently did so also. No matter which figure is correct, there is still a disparate impact.
The Court quotes Teamsters v. United States, 431 U. S. 324, 340 n. 20 (1977), to the effect that “‘evidence showing that the figures for the general population might not accurately reflect the pool of qualified job applicants’ undermines the significance of such figures.” Ante, at 586, n. 29. Petitioners have not put on such “evidence”; we have only the Court’s hypotheses, facially unlikely ones at that. Under the Federal Rules of Evidence, to be admissibly relevant, evidence must only tend to establish a material fact. This evidence does that, and by definition unrebutted probative evidence on the material fact is sufficient to make out a prima facie case.
Petitioners suggest that the evidence did not include private clinics since the Center does not receive information from them. Had this objection been raised in the District Court as it should have been, respondents would have had the opportunity to remove any doubt about whether the evidence included private programs. Moreover, in support of their suggestion, petitioners rely upon two isolated statements that do not directly discuss the study in question. Dr. Lukoff testified that the private clinics report to the FDA but not to the “Rockefeller Institute register,” and he estimated that there were about 1,500 patients in such unreporting clinics. Tr. 252 (Jan. 9, 1975) (emphasis supplied). Dr. Dole, a professor at Rockefefier University and senior physician at the University Hospital, testified that “the methadone data center . . . maintains the computerized inventory on aü 40,000 patients in treatment” and that “[a]ll of the known programs report, I presume.” Id., at 114 (Jan. 7, 1975) (emphasis supplied). He did testify that “[tjhe most detañed documentation comes from the major public” programs, which “comprise about 25,000 out of the 40,000” methadone patients. As to the remaining patients, his program stiff had “simpl[e] registry information . . . .” Id., at 115-116. In short, the majority’s unsupported effort to undermine the District Court’s findings of fact merely establishes the wisdom of either remanding or, on the Court’s evident assumption that the Court of Appeals would have affirmed the Title VII judgment, abiding by the “two-court rule.”
The evidence before the District Court established that 80% of heroin addicts in the New York City metropolitan area, the source of clients for both public and private methadone clinics, are black or Hispanic.
Because the rule is unwritten in relevant part, there is confusion about its scope. The Court asserts that it does not exclude those who formerly used methadone, and that the District Court “did not settle the question of what policy TA enforces in this respect . . . .” Ante, at 572 n. 3. In fact, however, petitioners openly admit that they automatically exclude former methadone users unless they “have been completely drug free and have had a stable history for at least five years.” Brief for Petitioners 5. And I quote the District Court’s actual finding which in context is unlike that described by the majority:
“It is clear that a relatively recent methadone user would be subject to the blanket exclusionary policy. However, the TA has indicated that there might be some flexibility with respect to a person who had once used methadone, but had been free of such use for a period of five years or more.” 399 F. Supp., at 1036.
The Court finds no “concrete dispute between adverse litigants” over the former-users policy because no former user is entitled to relief under the District Court’s injunction. Ante, at 573 n. 3. But respondent Frasier is a former user, see ante, at 576-577, n. 12, and the District Court expressly granted him relief, including backpay from the time he was rejected as a recent former methadone user. App. to Pet. for Cert. 77a-78a. The Court says the District Court erred in finding as facts that Frasier was using no narcotics in April 1973 and that petitioners refused to hire him solely because of his prior, apparently successful methadone treatment. As I read the facts as recited by the Court, the District Court was clearly correct, but in any event petitioners have not preserved this argument in the Court of Appeals or here. See Defendants’ Proposed Findings of Fact 6-7 (filed Oct. 18, 1974) (Frasier “purportedly” graduated successfully from the methadone program on March 19, 1973, and, though otherwise eligible, was rejected due to “his drug history” on April 2, 1973). See also ante, at 596 n. 4 (Powell, J., dissenting in relevant part).
The Court apparently reads the District Court’s injunction as protecting only those persons who had been in methadone programs for a year or longer before they were cured. It is incredible that the District Court would have punished those persons able to triumph over heroin addiction in less than a year. And the context of the District Court’s order, combined with the grant of relief to respondent Frasier, makes it clear that the court intended to protect, and had good reason to do so, all former *604methadone users as well as those current users who have been successfully maintained for more than a year.
Respondents presented numerous top experts in this field and large employers experienced with former heroin users treated with methadone. Both sides rested after six days of trial, but the District Court demanded nine more days of further factual development, and an 8-hour inspection of petitioners’ facilities, because it did not believe that the evidence could be so one-sidedly in respondents’ favor. The court correctly realized its responsibility in a public-law case of this type to demand the whole story before making a constitutional ruling. See Chayes, The Role of the Judge in Public Law Litigation, 89 Harv. L. Rev. 1281 (1976). The District Court called six witnesses of its own, and it chose them primarily because they had written articles on methadone maintenance that petitioners asserted had shown the unreliability of that method of dealing with heroin addiction. It also correctly expressed its refusal to base its judgment on shifting medical opinions.
The rule’s treatment of those who succeed is at issue here, since the District Court effectively amended the complaint to allege discrimination against that subgroup, see Fed. Rule Civ. Proc. 15 (b), and implicitly found no constitutional violation with respect to others burdened by the practice.
The evidence indicates that poor risks will shake out of a methadone maintenance program within six months. 399 F. Supp., at 1048-1049. It is a measure of the District Court’s caution that it set a 1-year standard.
A major sponsor of the recent amendments to the Rehabilitation Act, see ante, at 580-581, and n. 16, described the congressional determination behind them as being that a public employer “cannot assume that a history of alcoholism or drug addiction, including a past addiction currently treated by methadone maintenance, poses Sufficient danger in and of itself to justify exclusion [from employment]. Such an assumption would have no basis in fact . . . .” 124 Cong. Rec. 37510 (1978) (Sen. Williams).
Though a person free of illicit drug use for one year might subsequently revert, those who have graduated from methadone programs might do so also, and the Court apparently believes that the employment exclusion could not constitutionally be extended to them. See ante, at 572-573, n. 3, and 591-592, n. 37. See also ante, at 596-597 (Powell, J., dissenting in relevant part).
Since the District Court found as a fact that the bad risks could be culled from this group through the normal processing of employment applications, the only possible justification for this rule is that it eliminates applications in which petitioners would invest some time and effort before finding the person unemployable. The problem, however, is that not everyone in the general population is employable. Thus, if vacancies are to be filled, individualized hiring decisions must be made in any event.
The fact of methadone use must be determined somehow, so all applications must at least be read, and petitioners require all applicants under 35, and many existing employees, to submit to urinalysis. Reading the applications may disclose not only the fact of methadone use but also whether the person has certain educational or other qualifications and whether he or she has had a stable employment experience or any recent job-related difficulties.
The Court says that petitioners would be burdened by having to verify that a methadone applicant was successful in his program. But the program itself verifies that fact, and the District Court found that all petitioners would have to do is get in touch with the program, and that “this is essentially no different from obtaining relevant references for other types of applicants.” 399 F. Supp., at 1050 n. "3. A number of expert witnesses testified that the methadone clinics have far more information about their patients than personnel officers could ordinarily hope to acquire. The Court fears that some of the programs might not be reliable, but the District Court found that most are and ruled that petitioners do not have to hire any applicant “where there is reason to doubt the reliability of” the information furnished by the applicant’s clinic. Id., at 1058; accord, id., at 1050 n. 3. Consequently, I see no error at all, much less clear error, in the District Court’s finding of fact that petitioners “can perform this screening for methadone maintenance patients in basically the same way as in the case of other prospective employees.” Id., at 1048; accord, id., at 1037 and 1050 n. 3.
As to supervision of those who are hired, the fact that they present no greater risk than any other employee eliminates the need for any special supervision, except perhaps a notation on their personnel files that they *609need not be assigned to safety-sensitive positions. The District Court found as a fact that petitioners’ methods of monitoring all their employees “can be used for persons on methadone maintenance just as they are used for other persons . . . .” Id., at 1037.
I have difficulty also with the Court’s easy conclusion that the challenged rule was “[q]uite plainly” not motivated “by any special animus against a specific group of persons.” Ante, at 593 n. 40. Heroin addiction is a special problem of the poor, and the addict population is composed largely of racial minorities that the Court has previously recognized as politically powerless and historical subjects of majoritarian neglect. Persons on methadone maintenance have few interests in common with members of the majority, and thus are unlikely to have their interests protected, or even considered, in governmental decisionmaking. Indeed, petitioners stipulated that “[o]ne of the reasons for the . . . drug policy is the fact that [petitioners] fee[l] an adverse public reaction would result if it were generally known that [petitioners] employed persons with a prior history of drug abuse, including persons participating in methadone maintenance programs.” App. 83A. It is hard for me to reconcile that stipulation of animus against former addicts with our past holdings that “a bare . . . desire to harm a politically unpopular group cannot constitute a legitimate governmental interest.” United States Dept. of Agriculture v. Moreno, 413 U. S. 528, 534 (1973). On the other hand, the afflictions to which petitioners are more sympathetic, such as alcoholism and mental illness, are shared by both white and black, rich and poor.
Some weight should also be given to the history of the rule. See Arlington Heights v. Metropolitan Housing Dev. Corp., 429 U. S. 252, 267-268 (1977). Petitioners admit that it was not the result of a reasoned policy decision and stipulated that they had never studied the ability of those on methadone maintenance to perform petitioners’ jobs. Petitioners are not directly accountable to the public, are not the type of official body that normally makes legislative judgments of fact such as those relied upon by the majority today, and are by nature more concerned with business efficiency than with other public policies for which they have no direct responsibility. Cf. Hampton v. Mow Sun Wong, 426 U. S. 88, 103, (1976). But see ante, at 592. Both the State and City of New York, which do *610exhibit those democratic characteristics, hire persons in methadone programs for similar jobs.
These factors together strongly point to a conclusion of invidious discrimination. The Court, however, refuses to view this rule as one “circum-scrib[ing] a class of persons characterized by some unpopular trait or affiliation,” ante, at 593, because it is admittedly justified as applied to many current and former heroin addicts. Because the challenged classification unfairly burdens only a portion of all heroin addicts, the Court reasons that it cannot possibly have been spurred by animus by the “ruling majority.” All that shows, however, is that the characteristic in question is a legitimate basis of distinction in some circumstances; heroin addiction is a serious affliction that will often affect employability. But sometimes antipathy extends beyond the facts that may have given rise to it, and when that happens the “stereotyped reaction may have no rational relationship— other than pure prejudicial discrimination — to the stated purpose for which the classification is being made.” Mathews v. Lucas, 427 U. S. 495, 520-521 (1976) (Stevens, J., dissenting; footnote omitted). That is the case here.
The District Court found that the only common physical effects of methadone maintenance are increases in sweating, insomnia, and constipation, and a decrease in sex drive. 399 F. Supp., at 1044^-1045. Those disabilities are unfortunate but are hardly related to inability to be a subway janitor. This Court hints that the employability of even those successfully being maintained on methadone might be reduced by their obligation to appear at their clinics three times a week. Ante, at 588-589, n. 32. But all employees have outside obligations, and petitioners have neither argued nor proved that this particular duty would interfere with work.
The District Court did find that a possible but rare effect of methadone is minor impairment of abilities “required for the performance of potentially hazardous tasks, such as driving a car or operating machinery,” 399 F. Supp., at 1045, and the court exempted from the relief ordered such positions as subway motorman, which require “unique sensitivity.” Id., at 1052. But this does not make rational the blanket exclusion from all jobs, regardless of the qualifications required.
The District Court found, and petitioners have not challenged, that current problem drinkers present more of an employment risk than do respondents. Petitioners do not automatically discharge employees who are found to have a drinking problem. Id., at 1058.
The Court argues that “the fact that [petitioners have] the resources to expend on one class of problem employees does not by itself establish a constitutional duty on [their] part to come up with resources to spend on all classes of problem employees.” Ante, at 591-592, n. 37. If respondents were demanding to have the benefit of a rehabilitation program extended to them, petitioners could perhaps argue for freedom to deal with only one problem at a time due to limited resources. See Williamson v. Lee Optical Co., 348 U. S. 483, 489 (1955). In that situation, the lack of resources, or the desire to experiment in a limited field, might be a legitimate objective explaining the classification. But respondents are not asking for special, beneficial treatment; they are asking why they should be absolutely excluded from the opportunity to compete for petitioners’ jobs.