Justice Stevens

delivered the opinion of the Court.

The petitioner, Western Air Lines, Inc., requires that its flight engineers retire at age 60. Although the Age Discrimination in Employment Act of 1967 (ADEA), 29 U. S. C. *403§§621-634, generally prohibits mandatory retirement before age 70, the Act provides an exception “where age is a bona fide occupational qualification [BFOQ] reasonably necessary to the normal operation of the particular business.”1 A jury concluded that Western’s mandatory retirement rule did not qualify as a BFOQ even though it purportedly was adopted for safety reasons. The question here is whether the jury was properly instructed on the elements of the BFOQ defense.2

I

In its commercial airline operations, Western operates a variety of aircraft, including the Boeing 727 and the McDonnell-Douglas DC-10. These aircraft require three crew members in the cockpit: a captain, a first officer, and a flight engineer. “The ‘captain’ is the pilot and controls the aircraft. He is responsible for all phases of its operation. The ‘first officer’ is the copilot and assists the captain. The ‘flight engineer’ usually monitors a side-facing instrument panel. He does not operate the flight controls unless the captain and the first officer become incapacitated.” Trans World Airlines, Inc. v. Thurston, 469 U. S. 111, 114 (1985).

*404A regulation of the Federal Aviation Administration (FAA) prohibits any person from serving as a pilot or first officer on a commercial flight “if that person has reached his 60th birthday.” 14 CFR § 121.383(c) (1985). The FAA has justified the retention of mandatory retirement for pilots on the theory that “incapacitating medical events” and “adverse psychological, emotional, and physicial changes” occur as a consequence of aging. “The inability to detect or predict with precision an individual’s risk of sudden or subtle incapacitation, in the face of known age-related risks, counsels against relaxation of the rule.” 49 Fed. Reg. 14695 (1984). See also 24 Fed. Reg. 9776 (1959).

At the same time, the FAA has refused to establish a mandatory retirement age for flight engineers. “While a flight engineer has important duties which contribute to the safe operation of the airplane, he or she may not assume the responsibilities of the pilot in command.” 49 Fed. Reg., at 14694. Moreover, available statistics establish that flight engineers have rarely been a contributing cause or factor in commercial aircraft “accidents” or “incidents.” Ibid.

In 1978, respondents Criswell and Starley were captains operating DC-10s for Western. Both men celebrated their 60th birthdays in July 1978. Under the collective-bargaining agreement in effect between Western and the union, cockpit crew members could obtain open positions by bidding in order of seniority.3 In order to avoid mandatory retirement *405under the FAA’s under-age-60 rule for pilots, Criswell and Starley applied for reassignment as flight engineers. Western denied both requests, ostensibly on the ground that both employees were members of the company’s retirement plan which required all crew members to retire at age 60.4 For the same reason, respondent Ron, a career flight engineer, was also retired in 1978 after his 60th birthday.

Mandatory retirement provisions similar to those contained in Western’s pension plan had previously been upheld under the ADEA. United Air Lines, Inc. v. McMann, 434 U. S. 192 (1977). As originally enacted in 1967, the Act provided an exception to its general proscription of age discrimination for any actions undertaken “to observe the terms of a. . . bona fide employee benefit plan such as a retirement, pension, or insurance plan, which is not a subterfuge to evade the purposes of this Act.”5 In April 1978, however, Congress amended the statute to prohibit employee benefit plans from requiring the involuntary retirement of any employee because of age.6

Criswell, Starley, and Ron brought this action against Western contending that the under-age-60 qualification for *406the position of flight engineer violated the ADEA. In the District Court, Western defended, in part, on the theory that the age-60 rule is a BFOQ “reasonably necessary” to the safe operation of the airline.7 All parties submitted evidence concerning the nature of the flight engineer’s tasks, the physiological and psychological traits required to perform them, and the availability of those traits among persons over age 60.

As the District Court summarized, the evidence at trial established that the flight engineer’s “normal duties are less critical to the safety of flight than those of a pilot.” 514 F. Supp. 384, 390 (CD Cal. 1981). The flight engineer, however, does have critical functions in emergency situations and, of course, might cause considerable disruption in the event of his own medical emergency.

The actual capabilities of persons over age 60, and the ability to detect disease or a precipitous decline in their faculties, were the subject of conflicting medical testimony. Western’s expert witness, a former FAA Deputy Federal Air Surgeon,8 was especially concerned about the possibility of a “cardiovascular event” such as a heart attack. He testified that “with advancing age the likelihood of onset of disease increases and that in persons over age 60 it could not be predicted whether and when such diseases would occur.” Id., at 389.

The plaintiffs’ experts, on the other hand, testified that physiological deterioration is caused by disease, not aging, and that “it was feasible to determine on the basis of individual medical examinations whether flight deck crew members, including those over age 60, were physically qualified to con*407tinue to fly.” Ibid. These conclusions were corroborated by the nonmedical evidence:

“The record also reveals that both the FAA and the airlines have been able to deal with the health problems of pilots on an individualized basis. Pilots who have been grounded because of alcoholism or cardiovascular disease have been recertified by the FAA and allowed to resume flying. Pilots who were unable to pass the necessary examination to maintain their FAA first class medical certificates, but who continued to qualify for second class medical certificates were allowed to ‘downgrade’ from pilot to [flight engineer]. There is nothing in the record to indicate that these flight deck crew members are physically better able to perform their duties than flight engineers over age 60 who have not experienced such events or that they are less likely to become incapacitated.” Id., at 390.

Moreover, several large commercial airlines have flight engineers over age 60 “flying the line” without any reduction in their safety record. Ibid.

The jury was instructed that the “BFOQ defense is available only if it is reasonably necessary to the normal operation or essence of defendant’s business.” Tr. 2626. The jury was informed that “the essence of Western’s business is the safe transportation of their passengers.” Ibid. The jury was also instructed:

“One method by which defendant Western may establish a BFOQ in this case is to prove:

“(1) That in 1978, when these plaintiffs were retired, it was highly impractical for Western to deal with each second officer over age 60 on an individualized basis to determine his particular ability to perform his job safely; and

“(2) That some second officers over age 60 possess traits of a physiological, psychological or other nature *408which preclude safe and efficient job performance that cannot be ascertained by means other than knowing their age.

“In evaluating the practicability to defendant Western of dealing with second officers over age 60 on an individualized basis, with respect to the medical testimony, you should consider the state of the medical art as it existed in July 1978.” Id., at 2627.

The jury rendered a verdict for the plaintiffs, and awarded damages. After trial, the District Court granted equitable relief, explaining in a written opinion why it found no merit in Western’s BFOQ defense to the mandatory retirement rule. 514 F. Supp., at 389-391.9

On appeal, Western made various arguments attacking the verdict and judgment below, but the Court of Appeals affirmed in all respects. 709 F. 2d 544 (CA9 1983). In particular, the Court of Appeals rejected Western’s contention that the instruction on the BFOQ defense was insufficiently deferential to the airline’s legitimate concern for the safety of its passengers. Id., at 549-551. We granted certiorari to consider the merits of this question. 469 U. S. 815 (1984).10

*409h-i hH

Throughout the legislative history of the ADEA, one empirical fact is repeatedly emphasized: the process of psychological and physiological degeneration caused by aging varies with each individual. “The basic research in the field of aging has established that there is a wide range of individual physical ability regardless of age.”11 As a result, many older American workers perform at levels equal or superior to their younger colleagues.

In 1965, the Secretary of Labor reported to Congress that despite these well-established medical facts there “is persistent and widespread use of age limits in hiring that in a great many cases can be attributed only to arbitrary discrimination against older workers on the basis of age and regardless of ability.”12 Two years later, the President recommended that Congress enact legislation to abolish arbitrary age limits on *410hiring. Such limits, the President declared, have a devastating effect on the dignity of the individual and result in a staggering loss of human resources vital to the national economy.13

After further study,14 Congress responded with the enactment of the ADEA. The preamble declares that the purpose of the ADEA is “to promote employment of older persons based on their ability rather than age [and] to prohibit arbitrary age discrimination in employment.” 81 Stat. 602, 29 U. S. C. § 621(b). Section 4(a)(1) makes it “unlawful for an employer ... to fail or refuse to hire or to discharge any individual or otherwise discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s age.” 81 Stat. 603, 29 U. S. C. § 623(a)(1). This proscription presently applies to all persons between the ages of 40 and 70. 29 U. S. C. § 631(a).

The legislative history of the 1978 Amendments to the ADEA makes quite clear that the policies and substantive provisions of the Act apply with especial force in the case of mandatory retirement provisions. The House Committee on Education and Labor reported:

“Increasingly, it is being recognized that mandatory retirement based solely upon age is arbitrary and that chronological age alone is a poor indicator of ability to perform a job. Mandatory retirement does not take *411into consideration actual differing abilities and capacities. Such forced retirement can cause hardships for older persons through loss of roles and loss of income. Those older persons who wish to be re-employed have a much more difficult time finding a new job than younger persons.

“Society, as a whole, suffers from mandatory retirement as well. As a result of mandatory retirement, skills and experience are lost from the work force resulting in reduced GNP. Such practices also add a burden to Government income maintenance programs such as social security.”15

In the 1978 Amendments, Congress narrowed an exception to the ADEA which had previously authorized involuntary retirement under limited circumstances. See supra, at 405.

In both 1967 and 1978, however, Congress recognized that classifications based on age, like classifications based on religion, sex, or national origin, may sometimes serve as a necessary proxy for neutral employment qualifications essential to the employer’s business. The diverse employment situations in various industries, however, forced Congress to adopt a “case-by-case basis ... as the underlying rule in the administration of the legislation.” H. R. Rep. No. 805, 90th Cong., 1st Sess., 7 (1967), Legislative History 80.16 Congress offered only general guidance on when an age clas*412sification might be permissible by borrowing a concept and statutory language from Title VII of the Civil Rights Act of 196417 and providing that such a classification is lawful “where age is a bona fide occupational qualification reasonably necessary to the normal operation of the particular business.” 29 U. S. C. § 623(f)(1).

Shortly after the passage of the Act, the Secretary of Labor, who was at that time charged with its enforcement, adopted regulations declaring that the BFOQ exception to the ADEA has only “limited scope and application” and “must be construed narrowly.” 33 Fed. Reg. 9172 (1968), 29 CFR §860.102(b) (1984). The Equal Employment Opportunity Commission (EEOC) adopted the same narrow construction of the BFOQ exception after it was assigned authority for enforcing the statute. 46 Fed. Reg. 47727 (1981), 29 CFR § 1625.6 (1984). The restrictive language of the statute and the consistent interpretation of the administrative agencies charged with enforcing the statute convince us that, like its Title VII counterpart, the BFOQ exception “was in fact meant to be an extremely narrow exception to the general prohibition” of age discrimination contained in the ADEA. Dothard v. Rawlinson, 433 U. S. 321, 334 (1977).

HH 1 — 1 H-1

In Usery v. Tamiami Trail Tours, Inc., 531 F. 2d 224 (1976), the Court of Appeals for the Fifth Circuit was called upon to evaluate the merits of a BFOQ defense to a claim of age discrimination. Tamiami Trail Tours, Inc., had a policy of refusing to hire persons over age 40 as intercity bus drivers. At trial, the bus company introduced testimony supporting its theory that the hiring policy was a BFOQ based *413upon safety considerations — the need to employ persons who have a low risk of accidents. In evaluating this contention, the Court of Appeals drew on its Title VII precedents, and concluded that two inquiries were relevant.

First, the court recognized that some job qualifications may be so peripheral to the central mission of the employer’s business that no age discrimination can be “reasonably necessary to the normal operation of the particular business.”18 29 U. S. C. § 623(f)(1). The bus company justified the age qualification for hiring its drivers on safety considerations, but the court concluded that this claim was to be evaluated under an objective standard:

“[T]he job qualifications which the employer invokes to justify his discrimination must be reasonably necessary to the essence of his business — here, the safe transportation of bus passengers from one point to another. The greater the safety factor, measured by the likelihood of harm and the probable severity of that harm in case of an accident, the more stringent may be the job qualifications designed to insure safe driving.” 531 F. 2d, at 236.

This inquiry “adjusts to the safety factor” by ensuring that the employer’s restrictive job qualifications are “reasonably necessary” to further the overriding interest in public safety. Ibid. In Tamiami, the court noted that no one had seriously *414challenged the bus company’s safety justification for hiring drivers with a low risk of having accidents.

Second, the court recognized that the ADEA requires that age qualifications be something more than “convenient” or “reasonable”; they must be “reasonably necessary ... to the particular business,” and this is only so when the employer is compelled to rely on age as a proxy for the safety-related job qualifications validated in the first inquiry.19 This showing could be made in two ways. The employer could establish that it “‘had reasonable cause to believe, that is, a factual basis for believing, that all or substantially all [persons over the age qualifications] would be unable to perform safely and efficiently the duties of the job involved.’”20 In Tamiami, the employer did not seek to justify its hiring qualification under this standard.

Alternatively, the employer could establish that age was a legitimate proxy for the safety-related job qualifications by proving that it is “ ‘impossible or highly impractical’ ” to deal with the older employees on an individualized basis.21 “One method by which the employer can carry this burden is to establish that some members of the discriminated-against class possess a trait precluding safe and efficient job performance *415that cannot be ascertained by means other than knowledge of the applicant’s membership in the class.” Id., at 235. In Tamiami, the medical evidence on this point was conflicting, but the District Court had found that individual examinations could not determine which individuals over the age of 40 would be unable to operate the buses safely. The Court of Appeals found that this finding of fact was not “clearly erroneous,” and affirmed the District Court’s judgment for the bus company on the BFOQ defense. Id., at 238.

Congress, in considering the 1978 Amendments, implicitly endorsed the two-part inquiry identified by the Fifth Circuit in the Tamiami case. The Senate Committee Report expressed concern that the amendment prohibiting mandatory retirement in accordance with pension plans might imply that mandatory retirement could not be a BFOQ:

“For example, in certain types of particularly arduous law enforcement activity, there may be a factual basis for believing that substantially all employees above a specified age would be unable to continue to perform safely and efficiently the duties of their particular jobs, and it may be impossible or impractical to determine through medical examinations, periodic reviews of current job performance and other objective tests the employees’ capacity or ability to continue to perform the jobs safely and efficiently.

“Accordingly, the committee adopted an amendment to make it clear that where these two conditions are satisfied and where such a bona fide occupational qualification has therefore been established, an employer may lawfully require mandatory retirement at that specified age.” S. Rep. No. 95-493, pp. 10-11 (1977), Legislative History 443-444.

The amendment was adopted by the Senate, but deleted by the Conference Committee because it “neither added to nor *416worked any change upon present law.”22 H. R. Conf. Rep. No. 95-950, p. 7 (1978), Legislative History 518.

Every Court of Appeals that has confronted a BFOQ defense based on safety considerations has analyzed the problem consistently with the Tamiami standard.23 An EEOC regulation embraces the same criteria.24 Considering the narrow language of the BFOQ exception, the parallel treatment of such questions under Title VII, and the uniform application of the standard by the federal courts, the EEOC, and Congress, we conclude that this two-part inquiry prop*417erly identifies the relevant considerations for resolving a BFOQ defense to an age-based qualification purportedly justified by considerations of safety.

hH

In the trial court, Western preserved an objection to any instruction in the Tamiami mold, claiming that “any instruction pertaining to the statutory phrase ‘reasonably necessary to the normal operation of [defendant’s] business’... is irrelevant to and confusing for the deliberations of the jury.”25 Western proposed an instruction that would have allowed it to succeed on the BFOQ defense by proving that “in 1978, when these plaintiffs were retired, there existed a rational basis in fact for defendant to believe that use of [flight engineers] over age 60 on its DC-10 airliners would increase the likelihood of risk to its passengers.”26 The proposed instruction went on to note that the jury might rely on the FAA’s age-60 rule for pilots to establish a BFOQ under this standard “without considering any other evidence.”27 It also noted that the medical evidence submitted by the parties might provide a “rational basis in fact.”

On appeal, Western defended its proposed instruction, and the Court of Appeals soundly rejected it. 709 F. 2d, at • 549-551. In this Court, Western slightly changes its course. *418The airline now acknowledges that the Tamiami standard identifies the relevant general inquiries that must be made in evaluating the BFOQ defense. However, Western claims that in several respects the instructions given below were insufficiently protective of public safety. Western urges that we interpret or modify the Tamiami standard to weigh these concerns in the balance.

Reasonably Necessary Job Qualifications

Western relied on two different kinds of job qualifications to justify its mandatory retirement policy. First, it argued that flight engineers should have a low risk of incapacitation or psychological and physiological deterioration. At this vague level of analysis respondents have not seriously disputed — nor could they — that the qualification of good health for a vital crew member is reasonably necessary to the essence of the airline’s operations. Instead, they have argued that age is not a necessary proxy for that qualification.

On a more specific level, Western argues that flight engineers must meet the same stringent qualifications as pilots, and that it was therefore quite logical to extend to flight engineers the FAA’s age-60 retirement rule for pilots. Although the FAA’s rule for pilots, adopted for safety reasons, is relevant evidence in the airline’s BFOQ defense, it is not to be accorded conclusive weight. Johnson v. Mayor and City Council of Baltimore, ante, at 370-371. The extent to which the rule is probative varies with the weight of the evidence supporting its safety rationale and “the congruity between the . . . occupations at issue.” Ante, at 371. In this case, the evidence clearly established that the FA A, Western, and other airlines all recognized that the qualifications for a flight engineer were less rigorous than those required for a pilot.28

*419In the absence of persuasive evidence supporting its position, Western nevertheless argues that the jury should have been instructed to defer to “Western’s selection of job qualifications for the position of [flight engineer] that are reasonable in light of the safety risks.” Brief for Petitioner 30. This proposal is plainly at odds with Congress’ decision, in adopting the ADEA, to subject such management decisions to a test of objective justification in a court of law. The BFOQ standard adopted in the statute is one of “reasonable necessity,” not reasonableness.

In adopting that standard, Congress did not ignore the public interest in safety. That interest is adequately reflected in instructions that track the language of the statute. When an employer establishes that a job qualification has been carefully formulated to respond to documented concerns for public safety, it will not be overly burdensome to persuade a trier of fact that the qualification is “reasonably necessary” to safe operation of the business. The uncertainty implicit in the concept of managing safety risks always makes it “reasonably necessary” to err on the side of caution in a close case.29 The employer cannot be expected to establish the risk of an airline accident “to a certainty, for certainty would require running the risk until a tragic accident would *420prove that the judgment was sound.” Usery v. Tamiami Trail Tours, Inc., 531 F. 2d, at 238. When the employer’s argument has a credible basis in the record, it is difficult to believe that a jury of laypersons — many of whom no doubt have flown or could expect to fly on commercial air carriers— would not defer in a close case to the airline’s judgment. Since the instructions in this case would not have prevented the airline from raising this contention to the jury in closing argument, we are satisfied that the verdict is a consequence of a defect in Western’s proof rather than a defect in the trial court’s instructions.30

Western’s Statutory Safety Obligation

The instructions defined the essence of Western’s business as “the safe transportation of their passengers.” Tr. 2626. Western complains that this instruction was defective because it failed to inform the jury that an airline must conduct its operations “with the highest possible degree of safety.”31

Jury instructions, of course, “may not be judged in artificial isolation,” but must be judged in the “context of the overall charge” and the circumstances of the case. See Cupp v. Naughten, 414 U. S. 141, 147 (1973). In this case, the instructions characterized safe transportation as the “essence” *421of Western’s business and specifically referred to the importance of “safe and efficient job performance” by flight engineers. Tr. 2627. Moreover, in closing argument counsel pointed out that because “safety is the essence of Western’s business,” the airline strives for “the highest degree possible of safety.”32 Viewing the record as a whole, we are satisfied that the jury’s attention was adequately focused on the importance of safety to the operation of Western’s business. Cf. United States v. Park, 421 U. S. 658, 674 (1975).

Age as a Proxy for Job Qualifications

Western contended below that the ADEA only requires that the employer establish “a rational basis in fact” for believing that identification of those persons lacking suitable qualifications cannot occur on an individualized basis.33 This “rational basis in fact” standard would have been tantamount to an instruction to return a verdict in the defendant’s favor. Because that standard conveys a meaning that is significantly different from that conveyed by the statutory phrase “reasonably necessary,” it was correctly rejected by the trial court.34

*422Western argues that a “rational basis” standard should be adopted because medical disputes can never be proved “to a certainty” and because juries should not be permitted “to resolve bona fide conflicts among medical experts respecting the adequacy of individualized testing.” Reply Brief for Petitioner 9, n. 10. The jury, however, need not be convinced beyond all doubt that medical testing is impossible, but only that the proposition is true “on a preponderance of the evidence.” Moreover, Western’s attack on the wisdom of assigning the resolution of complex questions to 12 laypersons is inconsistent with the structure of the ADEA. Congress expressly decided that problems involving age discrimination in employment should be resolved on a “case-by-case basis” by proof to a jury.35

The “rational basis” standard is also inconsistent with the preference for individual evaluation expressed in the language and legislative history of the ADEA.36 Under the Act, employers are to evaluate employees between the ages of 40 and 70 on their merits and not their age. In the BFOQ defense, Congress provided a limited exception to this general principle, but required that employers validate any discrimination as “reasonably necessary to the normal operation of the particular business.” It might well be “rational” to require mandatory retirement at any age less than 70, but that result would not comply with Congress’ direction that employers must justify the rationale for the age chosen. Unless an employer can establish a substantial basis for believing that all or nearly all employees above an age lack the qualifications required for the position, the age selected for mandatory retirement less than 70 must be an age at which it *423is highly impractical for the employer to insure by individual testing that its employees will have the necessary qualifications for the job.

Western argues that its lenient standard is necessary because “where qualified experts disagree as to whether persons over a certain age can be dealt with on an individual basis, an employer must be allowed to resolve that controversy in a conservative manner.” Reply Brief for Petitioner 8-9. This argument incorrectly assumes that all expert opinion is entitled to equal weight, and virtually ignores the function of the trier of fact in evaluating conflicting testimony. In this case, the jury may well have attached little weight to the testimony of Western’s expert witness. See supra, at 406, and n. 8. A rule that would require the jury to defer to the judgment of any expert witness testifying for the employer, no matter how unpersuasive, would allow some employers to give free reign to the stereotype of older workers that Congress decried in the legislative history of the ADEA.

When an employee covered by the Act is able to point to reputable businesses in the same industry that choose to eschew reliance on mandatory retirement earlier than age 70, when the employer itself relies on individualized testing in similar circumstances, and when the administrative agency with primary responsibility for maintaining airline safety has determined that individualized testing is not impractical for the relevant position, the employer’s attempt to justify its decision on the basis of the contrary opinion of experts— solicited for the purposes of litigation — is hardly convincing on any objective standard short of complete deference. Even in cases involving public safety, the ADEA plainly does not permit the trier of fact to give complete deference to the employer’s decision.

The judgment of the Court of Appeals is

Affirmed.

Justice Powell took no part in the decision of this case.

Justice Stevens

announced the judgment of the Court and delivered the opinion of the Court with respect to Parts I, II, and IV, and an opinion with respect to Part III, in which Justice Souter, Justice Ginsburg, and Justice Breyer join.

Petitioners, police and public safety officers employed by the city of Jackson, Mississippi (hereinafter City), contend that salary increases received in 1999 violated the Age Discrimination in Employment Act of 1967 (ADEA) because they were less generous to officers over the age of 40 than to younger officers. Their suit raises the question whether the “disparate-impact” theory of recovery announced in Griggs v. Duke Power Co., 401 U. S. 424 (1971), for cases brought under Title VII of the Civil Rights Act of 1964, is cognizable under the ADEA. Despite the age of the ADEA, it is a question that we have not yet addressed. See Hazen *231 Paper Co. v. Biggins, 507 U. S. 604, 610 (1993); Markham v. Geller, 451 U. S. 945 (1981) (Rehnquist, J., dissenting from denial of certiorari).

I

On October 1, 1998, the City adopted a pay plan granting raises to all City employees. The stated purpose of the plan was to “attract and retain qualified people, provide incentive for performance, maintain competitiveness with other public sector agencies and ensure equitable compensation to all employees regardless of age, sex, race and/or disability.”1 On May 1, 1999, a revision of the plan, which was motivated, at least in part, by the City’s desire to bring the starting salaries of police officers up to the regional average, granted raises to all police officers and police dispatchers. Those who had less than five years of tenure received proportionately greater raises when compared to their former pay than those with more seniority. Although some officers over the age of 40 had less than five years of service, most of the older officers had more.

Petitioners are a group of older officers who filed suit under the ADEA claiming both that the City deliberately discriminated against them because of their age (the “disparate-treatment” claim) and that they were “adversely affected” by the plan because of their age (the “disparate-impact” claim). The District Court granted summary judgment to the City on both claims. The Court of Appeals held that the ruling on the former claim was premature because petitioners were entitled to further discovery on the issue of intent, but it affirmed the dismissal of the disparate-impact claim. 351 F. 3d 183 (CA5 2003). Over one judge’s dissent, the majority concluded that disparate-impact claims are categorically unavailable under the ADEA. Both the majority and the dissent assumed that the facts alleged by petitioners would entitle them to relief under the reasoning of Griggs.

*232We granted the officers’ petition for certiorari, 541 U. S. 958 (2004), and now hold that the ADEA does authorize recovery in “disparate-impact” cases comparable to Griggs. Because, however, we conclude that petitioners have not set forth a valid disparate-impact claim, we affirm.

II

During the deliberations that preceded the enactment of the Civil Rights Act of 1964, Congress considered and rejected proposed amendments that would have included older workers among the classes protected from employment discrimination.2 General Dynamics Land Systems, Inc. v. Cline, 540 U. S. 581, 587 (2004). Congress did, however, request the Secretary of Labor to “make a full and complete study of the factors which might tend to result in discrimination in employment because of age and of the consequences of such discrimination on the economy and individuals affected.” §715, 78 Stat. 265. The Secretary’s report, submitted in response to Congress’ request, noted that there was little discrimination arising from dislike or intolerance of older people, but that “arbitrary” discrimination did result from certain age limits. Report of the Secretary of Labor, The Older American Worker: Age Discrimination in Employment 5 (June 1965), reprinted in U. S. Equal Employment Opportunity Commission, Legislative History of the Age Discrimination in Employment Act (1981), Doc. No. 5 (hereinafter Wirtz Report). Moreover, the report observed that discriminatory effects resulted from “ Constitutional arrangements that indirectly restrict the employment of older workers.” Id., at 15.

In response to that report Congress directed the Secretary to propose remedial legislation, see Fair Labor Standards Amendments of 1966, Pub. L. 89-601, § 606, 80 Stat. 845, and *233then acted favorably on his proposal. As enacted in 1967, § 4(a)(2) of the ADEA, now codified as 29 U. S. C. § 623(a)(2), provided that it shall be unlawful for an employer “to limit, segregate, or classify his employees in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s age . . . .” 81 Stat. 603. Except for substitution of the word “age” for the words “race, color, religion, sex, or national origin,” the language of that provision in the ADEA is identical to that found in § 703(a)(2) of the Civil Rights Act of 1964 (Title VII). Other provisions of the ADEA also parallel the earlier statute.3 Unlike Title VII, however, § 4(f)(1) of the ADEA, 81 Stat. 603, contains language that significantly narrows its coverage by permitting any “otherwise prohibited” action “where the differentiation is based on reasonable factors other than age” (hereinafter RFOA provision).

III

In determining whether the ADEA authorizes disparate-impact claims, we begin with the premise that when Congress uses the same language in two statutes having similar purposes, particularly when one is enacted shortly after the other, it is appropriate to presume that Congress intended that text to have the same meaning in both statutes. Northcross v. Board of Ed. of Memphis City Schools, 412 U. S. 427, 428 (1973) (per curiam). We have consistently applied *234that presumption to language in the ADEA that was “derived in haec verba from Title VII.” Lorillard, v. Pons, 434 U. S. 575, 584 (1978).4 Our unanimous interpretation of § 703(a)(2) of Title VII in Griggs is therefore a precedent of compelling importance.

In Griggs, a case decided four years after the enactment of the ADEA, we considered whether § 703 of Title VII prohibited an employer “from requiring a high school education or passing of a standardized general intelligence test as a condition of employment in or transfer to jobs when (a) neither standard is shown to be significantly related to successful job performance, (b) both requirements operate to disqualify Negroes at a substantially higher rate than white applicants, and (c) the jobs in question formerly had been filled only by white employees as part of a longstanding practice of giving preference to whites.” 401 U. S., at 425-426. Accepting the Court of Appeals’ conclusion that the employer had adopted the diploma and test requirements without any intent to discriminate, we held that good faith “does not redeem employment procedures or testing mechanisms that operate as ‘built-in headwinds’ for minority groups and are unrelated to measuring job capability.” Id., at 432.

We explained that Congress had “directed the thrust of the Act to the consequences of employment practices, not simply the motivation.” Ibid. We relied on the fact that history is “filled with examples of men and women who rendered highly effective performance without the conventional badges of accomplishment in terms of certificates, diplomas, or degrees. Diplomas and tests are useful servants, but *235Congress has mandated the commonsense proposition that they are not to become masters of reality.” Id., at 433. And we noted that the Equal Employment Opportunity Commission (EEOC), which had enforcement responsibility, had issued guidelines that accorded with our view. Id., at 433-434. We thus squarely held that § 703(a)(2) of Title VII did not require a showing of discriminatory intent.5

While our opinion in Griggs relied primarily on the purposes of the Act, buttressed by the fact that the EEOC had endorsed the same view, we have subsequently noted that our holding represented the better reading of the statutory text as well. See Watson v. Fort Worth Bank & Trust, 487 U. S. 977, 991 (1988). Neither § 703(a)(2) nor the comparable language in the ADEA simply prohibits actions that “limit, segregate, or classify” persons; rather the language prohibits such actions that, “deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s” race or age. Ibid. (explaining that in disparate-impact cases, “the employer’s practices may be said to 'adversely affect [an individual’s status] as an employee’” (alteration in original) (quoting 42 *236U. S. C. § 2000e-2(a)(2))). Thus the text focuses on the effects of the action on the employee rather than the motivation for the action of the employer.6

Griggs, which interpreted the identical text at issue here, thus strongly suggests that a disparate-impact theory should be cognizable under the ADEA.7 Indeed, for over two dee-*237ades after our decision in Griggs, the Courts of Appeals uniformly interpreted the ADEA as authorizing recovery on a “disparate-impact” theory in appropriate cases.8 It was only after our decision in Hazen Paper Co. v. Biggins, 507 U. S. 604 (1993), that some of those courts concluded that the ADEA did not authorize a disparate-impact theory of liability.9 Our opinion in Hazen Paper, however, did not address or comment on the issue we decide today. In that case, we held that an employee’s allegation that he was discharged shortly before his pension would have vested did not state a cause of action under a disparate-treatment theory. The motivating factor was not, we held, the employee’s age, but rather his years of service, a factor that the ADEA did not prohibit an employer from considering when termi*238nating an employee. Id., at 612.10 While we noted that disparate treatment “captures the essence of what Congress sought to prohibit in the ADEA,” id., at 610, we were careful to explain that we were not deciding “whether a disparate impact theory of liability is available under the ADEA ... ibid. In sum, there is nothing in our opinion in Hazen Paper that precludes an interpretation of the ADEA that parallels our holding in Griggs.

The Court of Appeals’ categorical rejection of disparate-impact liability, like Justice O’Connor’s, rested primarily on the RFOA provision and the majority’s analysis of legislative history. As we have already explained, we think the history of the enactment of the ADEA, with particular reference to the Wirtz Report, supports the pre-Hazen Paper consensus concerning disparate-impact liability. And Hazen Paper itself contains the response to the concern over the RFOA provision.

The RFOA provision provides that it shall not be unlawful for an employer “to take any action otherwise prohibited under subsectio[n] (a). . . where the differentiation is based on reasonable factors other than age [discrimination] . . . .” 81 Stat. 603. In most disparate-treatment cases, if an employer in fact acted on a factor other than age, the action would not be prohibited under subsection (a) in the first place. See Hazen Paper, 507 U. S., at 609 (“[T]here is no disparate treatment under the ADEA when the factor motivating the employer is some feature other than the employee’s age”). In those disparate-treatment cases, such as in Hazen Paper itself, the RFOA provision is simply unnecessary to avoid liability under the ADEA, since there was no prohibited action in the first place. The RFOA provision is not, as Justice O’Connor suggests, a “safe harbor from liability,” post, at 252 (emphasis deleted), since there would *239be no liability under §4(a). See Texas Dept. of Community Affairs v. Burdine, 450 U. S. 248, 254 (1981) (noting, in a Title VII case, that an employer can defeat liability by showing that the employee was' rejected for “a legitimate, nondiscriminatory reason” without reference to an RFOA provision).

In disparate-impact cases, however, the allegedly “otherwise prohibited” activity is not based on age. Ibid. (“[C]laims that stress ‘disparate impact’ [by contrast] involve employment practices that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another...” (quoting Teamsters v. United States, 431 U. S. 324, 335-336, n. 15 (1977))). It is, accordingly, in cases involving disparate-impact claims that the RFOA provision plays its principal role by precluding liability if the adverse impact was attributable to. a nonage factor that was “reasonable.” Rather than support an argument that disparate impact is unavailable under the ADEA, the RFOA provision actually supports the contrary conclusion.11

Finally, we note that both the Department of Labor, which initially drafted the legislation, and the EEOC, which is the agency charged by Congress with responsibility for implementing the statute, 29 U. S. C. § 628, have consistently interpreted the ADEA to authorize relief on a disparate-impact theory. The initial regulations, while not mentioning disparate impact by' name, nevertheless permitted such claims if the employer relied on a factor that was not related to age. 29 CFR §860.103(f)(l)(i) (1970) (barring physical fitness requirements that were not “reasonably necessary for the spe-*240tifie work to be performed”). See also § 1625.7 (2004) (setting forth the standards for a disparate-impact claim).

The text of the statute, as interpreted in Griggs, the RFOA provision, and the EEOC regulations all support petitioners’ view. We therefore conclude that it was error for the Court of Appeals to hold that the disparate-impact theory of liability is categorically unavailable under the ADEA.

IV

Two textual differences between the ADEA and Title VII make it clear that even though both statutes authorize recovery on a disparate-impact theory, the scope of disparate-impact liability under ADEA is narrower than under Title VII. The first is the RFOA provision, which we have already identified. The second is the amendment to Title VII contained in the Civil Rights Act of 1991, 105 Stat. 1071. One of the purposes of that amendment was to modify the Court’s holding in Wards Cove Packing Co. v. Atonio, 490 U. S. 642 (1989), a case in which we narrowly construed the employer’s exposure to liability on a disparate-impact theory. See Civil Rights Act of 1991, § 2, 105 Stat. 1071. While the relevant 1991 amendments expanded the coverage of Title VII, they did not amend the ADEA or speak to the subject of age discrimination. Hence, Wards Cove’s pre-1991 interpretation of Title VII’s identical language remains applicable to the ADEA.

Congress’ decision to limit the coverage of the ADEA by including the RFOA provision is consistent with the fact that age, unlike race or other classifications protected by Title VII, not uncommonly has relevance to an individual’s capacity to engage in certain types of employment. To be sure, Congress recognized that this is not always the case, and that society may perceive those differences to be larger or more consequential than they are in fact. However, as Secretary Wirtz noted in his report, “certain circumstances . . . unquestionably affect older workers more strongly, as a *241group, than they do younger workers.” Wirtz Report 11. Thus, it is not surprising that certain employment criteria that are routinely used may be reasonable despite their adverse impact on older workers as a group. Moreover, intentional discrimination on the basis of age has not occurred at the same levels as discrimination against those protected by Title VII. While the ADEA reflects Congress’ intent to give older workers employment opportunities whenever possible, the RFOA provision reflects this historical difference.

Turning to the case before us, we initially note that petitioners have done little more than point out that the pay plan at issue is relatively less generous to older workers than to younger workers. They have not identified any specific test, requirement, or practice within the pay plan that has an adverse impact on older workers. As we held in Wards Cove, it is not enough to simply allege that there is a disparate impact on workers, or point to a generalized policy that leads to such an impact. Rather, the employee is “'responsible for isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities.’ ” 490 U. S., at 656 (quoting Watson, 487 U. S., at 994; emphasis added). Petitioners have failed to do so. Their failure to identify the specific practice being challenged is the sort of omission that could “result in employers being potentially liable for ‘the myriad of innocent causes that may lead to statistical imbalances . . . .’” 490 U. S., at 657. In this case not only did petitioners thus err by failing to identify the relevant practice, but it is also clear from the record that the City’s plan was based on reasonable factors other than age.

The plan divided each of five basic positions — police officer, master police officer, police sergeant, police lieutenant, and deputy police chief — into a series of steps and half-steps. The wage for each range was based on a survey of comparable communities in the Southeast. Employees were then assigned a step (or half-step) within their position that corres*242ponded to the lowest step that would still give the individual a 2% raise. Most of the officers were in the three lowest ranks; in each of those ranks there were officers under age 40 and officers over 40. In none did their age affect their compensation. The few officers in the two highest ranks are all over 40. Their raises, though higher in dollar amount than the raises given to junior officers, represented a smaller percentage of their salaries, which of course are higher than the salaries paid to their juniors. They are members of the class complaining of the “disparate impact” of the award.

Petitioners’ evidence established two principal facts: First, almost two-thirds (66.2%) of the officers under 40 received raises of more than 10% while less than half (45.3%) of those over 40 did.12 Second, the average percentage increase for the entire class of officers with less than five years of tenure was somewhat higher than the percentage for those with more seniority.13 Because older officers tended to occupy more senior positions, on average they received smaller increases when measured as a percentage of their salary. The basic explanation for the differential was the City’s perceived need to raise the salaries of junior officers to make them competitive with comparable positions in the market.

Thus, the disparate impact is attributable to the City’s decision to give raises based on seniority and position. Reliance on seniority and rank is unquestionably reasonable given the City’s goal of raising employees’ salaries to match those in surrounding communities. In sum, we hold that the City’s decision to grant a larger raise to lower echelon employees for the purpose of bringing salaries in line with that of surrounding police forces was a decision based on a “reasonable facto[r] other than age” that responded to the City’s legitimate goal of retaining police officers. Cf. MacPherson v. University of Montevallo, 922 F. 2d 766, 772 (CA11 1991).

*243While there may have been other reasonable ways for the City to achieve its goals, the one selected was not unreasonable. Unlike the business necessity test, which asks whether there are other ways for the employer to achieve its goals that do not result in a disparate impact on a protected class, the reasonableness inquiry includes no such requirement.

Accordingly, while we do not agree with the Court of Appeals’ holding that the disparate-impact theory of recovery is never available under the ADEA, we affirm its judgment.

It is so ordered.

The ChieF Justice took no part in the decision of this case.

Justice Scalia,

concurring in part and concurring in the judgment.

I concur in the judgment of the Court, and join all except Part III of its opinion. As to that Part, I agree with all of the Court’s reasoning, but would find it a basis, not for independent determination of the disparate-impact question, but for deferral to the reasonable views of the Equal Employment Opportunity Commission (EEOC or Commission) pursuant to Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837 (1984). See General Dynamics Land Systems, Inc. v. Cline, 540 U. S. 581, 601-602 (2004) (Scalia, J., dissenting).

This is an absolutely classic case for deference to agency interpretation. The Age Discrimination in Employment Act of 1967 (ADEA), 29 U. S. C. § 621 et seq., confers upon the EEOC authority to issue “such rules and regulations as it may consider necessary or appropriate for carrying out” the ADEA. §628. Pursuant to this authority, the EEOC promulgated, after notice-and-comment rulemaking, see 46 Fed. Reg. 47724, 47727 (1981), a regulation that reads as follows:

*244“When an employment practice, including a test, is claimed as a basis for different treatment of employees or applicants for employment on the grounds that it is a ‘factor other than’ age, and such a practice has an adverse impact on individuals within the protected age group, it can only be justified as a business necessity.” 29 CFR § 1625.7(d) (2004).

The statement of the EEOC which accompanied publication of the agency’s final interpretation of the ADEA said the following regarding this regulation: “Paragraph (d) of § 1625.7 has been rewritten to make it clear that employment criteria that are age-neutral on their face but which nevertheless have a disparate impact on members of the protected age group must be justified as a business necessity. See Laugesen v. Anaconda Corp., 510 F. 2d 307 (6th Cir. 1975); Griggs v. Duke Power Co., 401 U. S. 424 (1971).” 46 Fed. Reg., at 47725. The regulation affirmed, moreover, what had been the longstanding position of the Department of Labor, the agency that previously administered the ADEA, see ante, at 239; 29 CFR §860.103(f)(l)(i) (1970). And finally, the Commission has appeared in numerous cases in the lower courts, both as a party and as amicus curiae, to defend the position that the ADEA authorizes disparate-impact claims.1 Even under the unduly constrained standards of agency deference recited in United States v. Mead Corp., 533 *245U. S. 218 (2001), the EEOC’s reasonable view that the ADEA authorizes disparate-impact claims is deserving of deference. Id., at 229-231, and n. 12. A fortiori, it is entitled to deference under the pre-Mead formulation of Chevron, to which I continue to adhere. See 533 U. S., at 256-257 (Scalia, J., dissenting).

Justice O’Connor both denies that the EEOC has taken a position on the existence of disparate-impact claims and asserts that, even if it has, its position does not deserve deference. See post, at 264-267 (opinion concurring in judgment). The first claim cannot be squared with the text of the EEOC’s regulation, quoted above. This cannot possibly be read as agnostic on the question whether the ADEA prohibits employer practices that have a disparate impact on the aged. It provides that such practices “can only be justified as a business necessity,” compelling the conclusion that, absent a “business necessity,” such practices are prohibited.2

Justice O’Connor would not defer to the EEOC regulation, even if it read as it does, because, she says, the regulation “does not purport to interpret the language of § 4(a) at all,” but is rather limited to an interpretation of the “reasonable factors other than age” (RFOA) clause of § 4(f)(1) of the ADEA, which she says is not at issue. • Post, at 265. This argument assumes, however, that the RFOA clause operates independently of the remainder of the ADEA. It does not. Section 4(f)(1) provides, in relevant part:

*246“It shall not be unlawful for an employer, employment agency, or labor organization ... to take any action otherwise prohibited under subsections (a), (b), (c), or (e) of this section . . . where the differentiation is based on reasonable factors other than age . . . .” 29 U. S. C. § 623(f)(1) (emphasis added).

As this text makes clear, the RFOA defense is relevant only as a response to employer actions “otherwise prohibited” by the ADEA. Hence, the unavoidable meaning of the regulation at issue is that the ADEA prohibits employer actions that have an “adverse impact on individuals within the protected age group.” 29 CFR § 1625.7(d) (2004). And, of course, the only provision of the ADEA that could conceivably be interpreted to effect such a prohibition is § 4(a)(2)— the provision that Justice O’Connor maintains the EEOC “does not purport to interpret... at all.” Post, at 265.3

*247Lastly, Justice O’Connor argues that the EEOC’s interpretation of what is “otherwise prohibited” by the ADEA is not entitled to deference because the Court concludes that the same regulation’s interpretation of another term — the term “reasonable factors other than age,” which the regulation takes to include only “business necessity” — is unreasonable. Post, at 266. Her logic seems to be that, because the two interpretations appear in the same paragraph, they should stand or fall together. She cites no case for this proposition, and it makes little sense. If the two simultaneously adopted interpretations were contained in distinct paragraphs, the invalidation of one would not, of course, render the other infirm. (Justice O’Connor does not mean to imply, I assume, that our rejection of the EEOC’s application of the phrase “‘reasonable factors other than age’” to disparate-impact claims in paragraph (d) of § 1625.7 relieves the lower courts of the obligation to defer to the EEOC’s other applications of the same phrase in paragraph (c) or (e).) I can conceive no basis for a different rule simply because the two simultaneously adopted interpretations appear in the same paragraph.

The EEOC has express authority to promulgate rules and regulations interpreting the ADEA. It has exercised that authority to recognize disparate-impact claims. And, for the reasons given by the plurality opinion, its position is eminently reasonable. In my view, that is sufficient to resolve this case.

Justice O’Connor,

with whom Justice Kennedy and Justice Thomas join, concurring in the judgment.

“Disparate treatment . . . captures the essence of what Congress sought to prohibit in the [Age Discrimination in Employment Act of 1967 (ADEA), 29 U. S. C. § 621 et seq.] It is the very essence of age discrimination for an older employee to be fired because the employer believes that productivity and competence decline with old age.” Hazen Paper *248 Co. v. Biggins, 507 U. S. 604, 610 (1993). In the nearly four decades since the ADEA’s enactment, however, we have never read the statute to impose liability upon an employer without proof of discriminatory intent. See ibid.; Markham v. Geller, 451 U. S. 945 (1981) (Rehnquist, J., dissenting from denial of certiorari). I decline to join the Court in doing so today.

I would instead affirm the judgment below on the ground that disparate impact claims are not cognizable under the ADEA. The ADEA’s text, legislative history, and purposes together make clear that Congress did not intend the statute to authorize such claims. Moreover, the significant differences between the ADEA and Title VII of the Civil Rights Act of 1964 counsel against transposing to the former our construction of the latter in Griggs v. Duke Power Co., 401 U. S. 424 (1971). Finally, the agencies charged with administering the ADEA have never authoritatively construed the statute’s prohibitory language to impose disparate impact liability. Thus, on the precise question of statutory interpretation now before us, there is no reasoned agency reading of the text to which we might defer.

i

A

Our starting point is the statute’s text. Section 4(a) of the ADEA makes it unlawful for an employer:

“(1) to fail or refuse to hire or to discharge any individual or otherwise discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s age; [or]

“(2) to limit, segregate, or classify his employees in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s age . . . 29 U. S. C. § 623(a).

*249Neither petitioners nor the plurality contend that the first paragraph, § 4(a)(1), authorizes disparate impact claims, and I think it obvious that it does not. That provision plainly requires discriminatory intent, for to take an action against an individual “because of such individual’s age” is to do so “by reason of” or “on account of” her age. See Webster’s Third New International Dictionary 194 (1961); see also Teamsters v. United States, 431 U. S. 324, 335-336, n. 15 (1977) (“ ‘Disparate treatment’ ... is the most easily understood type of discrimination. The employer simply treats some people less favorably than others because of their [protected characteristic]. Proof of discriminatory motive is critical” (emphasis added)).

Petitioners look instead to the second paragraph, § 4(a)(2), as the basis for their disparate impact claim. But petitioners’ argument founders on the plain language of the statute, the natural reading of which requires proof of discriminatory intent. Section 4(a)(2) uses the phrase “because of... age” in precisely the same manner as does the preceding paragraph — to make plain that an employer is liable only if its adverse action against an individual is motivated by the individual’s age.

Paragraphs (a)(1) and (a)(2) do differ in one informative respect. The employer actions targeted by paragraph (a)(1) — i. e., refusing to hire, discharging, or discriminating against — are inherently harmful to the targeted individual. The actions referred to in paragraph (a)(2), on the other hand — i. e., limiting, segregating, or classifying — axe facially neutral. Accordingly, paragraph (a)(2) includes additional language which clarifies that, to give rise to liability, the employer’s action must actually injure someone: The decision to limit, segregate, or classify employees must “deprive or tend to deprive [an] individual of employment opportunities or otherwise adversely affect his status as an employee.” That distinction aside, the structures of paragraphs (a)(1) and (a)(2) are otherwise identical. Each paragraph prohibits an *250employer from taking specified adverse actions against an individual "because of such individual’s age.”

The plurality instead reads paragraph (a)(2) to prohibit employer actions that “adversely affect [an individual’s] status as an employe[e] because of such individual’s age.” Under this reading, “because of.. . age” refers to the cause of the adverse effect rather than the motive for the employer’s action. See ante, at 235-236. This reading is unpersuasive for two reasons. First, it ignores the obvious parallel between paragraphs (a)(1) and (a)(2) by giving the phrase “because of such individual’s age” a different meaning in each of the two paragraphs. And second, it ignores the drafters’ use of a comma separating the “because of . . . age” clause from the preceding language. That comma makes plain that the “because of . . . age” clause should not be read, as the plurality would have it, to modify only the “adversely affect” phrase. See, e. g., United States v. Ron Pair Enterprises, Inc., 489 U. S. 235, 241 (1989) (interpreting statute in light of the drafters’ use of a comma to set aside a particular phrase from the following language); see also B. Garner, A Dictionary of Modern Legal Usage 101 (2d ed. 1995) (“Generally, the word because should not follow a comma”). Rather, the “because of . . . age” clause is set aside to make clear that it modifies the entirety of the preceding paragraph: An employer may not, because of an individual’s age, limit, segregate, or classify his employees in a way that harms that individual.

The plurality also argues that' its reading is supported by the supposed “incongruity” between paragraph (a)(2)’s use of the plural in referring to the employer’s actions (“limit, segregate, or classify his employees”) and its use of the singular in the “because of such individual’s age” clause. (Emphases added.) Ante, at 236, n. 6. Not so. For the reasons just stated, the “because of. .. age” clause modifies all of the preceding language of paragraph (a)(2). That preceding language is phrased in both the plural (insofar as it *251refers to the employer’s actions relating to employees) and the singular (insofar as it requires that such action actually harm an individual). The use of the singular in the “because of . . . age” clause simply makes clear that paragraph (a)(2) forbids an employer to limit, segregate, or classify his employees if that decision is taken because of even one employee’s age and that individual (alone or together with others) is harmed.

B

While § 4(a)(2) of the ADEA makes it unlawful to intentionally discriminate because of age, § 4(f)(1) clarifies that “[i]t shall not be unlawful for an employer ... to take any action otherwise prohibited under subsections (a), (b), (c), or (e) of this section . . . where the differentiation is based on reasonable factors other than age . . . .” 29 U. S. C. § 623(f)(1). This “reasonable factors other than age” (RFOA) provision “insure[s] that employers [are] permitted to use neutral criteria” other than age, EEOC v. Wyoming, 460 U. S. 226, 232-233 (1983), even if this results in a disparate adverse impact on older workers. The provision therefore expresses Congress’ clear intention that employers not be subject to liability absent proof of intentional age-based discrimination. That policy, in my view, cannot easily be reconciled with the plurality’s expansive reading of § 4(a)(2).

The plurality, however, reasons that the RFOA provision’s language instead confirms that § 4(a) authorizes disparate impact claims. If § 4(a) prohibited only intentional discrimination, the argument goes, then the RFOA provision would have no effect because any action based on a factor other than age would not be “ ‘otherwise prohibited’ ” under § 4(a). See ante, at 238-239. Moreover, the plurality says, the RFOA provision applies only to employer actions based on reasonable factors other than age — so employers may still be held liable for actions based on treasonable nonage factors. See ante, at 239.

*252This argument misconstrues the purpose and effect of the RFOA provision. Discriminatory intent is required under §4(a), for the reasons discussed above. The role of the RFOA provision is to afford employers an independent safe harbor from liability. It provides that, where a plaintiff has made out a prima facie case of intentional age discrimination under §4(a) — thus “creating] a presumption that the employer unlawfully discriminated against the employee,” Texas Dept. of Community Affairs v. Burdine, 450 U. S. 248, 254 (1981) — the employer can rebut this case by producing evidence that its action was based on a reasonable nonage factor. Thus, the RFOA provision codifies a safe harbor analogous to the “legitimate, nondiscriminatory reason” (LNR) justification later recognized in Title VII suits. Ibid.; McDonnell Douglas Corp. v. Green, 411 U. S. 792, 802 (1973).

Assuming the McDonnell Douglas framework applies to ADEA suits, see O’Connor v. Consolidated Coin Caterers Corp., 517 U. S. 308, 311 (1996), this “rebuttal” function of the RFOA provision is arguably redundant with the judicially established LNR justification. See ante, at 238-239. But, at most, that merely demonstrates Congress’ abundance of caution in codifying an express statutory exemption from liability in the absence of discriminatory intent. See Fort Stewart Schools v. FLRA, 495 U. S. 641, 646 (1990) (provisions that, although “technically unnecessary,” are sometimes “inserted' out of an abundance of caution — a drafting imprecision venerable enough to have left its mark on legal Latin (ex obundanti cautela)”). It is noteworthy that even after McDonnell Douglas was decided, lower courts continued to rely on the RFOA exemption, in lieu of the LNR justification, as the basis for rebutting a prima facie case of age discrimination. See, e. g., Krieg v. Paul Revere Life Ins. Co., 718 F. 2d 998, 999 (CA11 1983) (per curiam); Schwager v. Sun Oil Co. of Pa., 591 F. 2d 58, 61 (CA10 1979); Bittar v. Air Canada, 512 F. 2d 582, 582-583 (CA5 1975) (per curiam).

*253In any event, the RFOA provision also plays a distinct (and clearly nonredundant) role in "mixed-motive" cases. In such cases, an adverse action taken in substantial part because of an employee's age may be "otherwise prohibited" by § 4(a). See Desert Palace, Inc. v. Costa, 539 U. S. 90, 93 (2003); Price Waterhouse v. Hopkins, 490 U. S. 228, 262-266 (1989) (O'Connor, J., concurring in judgment). The RFOA exemption makes clear that such conduct is nevertheless lawful so long as it is "based on" a reasonable factor other than age.

Finally, the RFOA provision's reference to "reasonable" factors serves only to prevent the employer from gaining the benefit of the statutory safe harbor by offering an irrational justification. Reliance on an unreasonable nonage factor would indicate that the employer's explanation is, in fact, no more than a pretext for intentional discrimination. See Reeves v. Sanderson Plumbing Products, Inc., 530 U. S. 133, 147 (2000); see also Hazen Paper, 507 U. S., at 613-614.

II

The legislative history of the ADEA confirms what its text plainly indicates-that Congress never intended the statute to authorize disparate impact claims. The drafters of the ADEA and the Congress that enacted it understood that age discrimination was qualitatively different from the kinds of discrimination addressed by Title VII, and that many legitimate employment practices would have a disparate impact on older workers. Accordingly, Congress determined that the disparate impact problem would best be addressed through noncoercive measures, and that the ADEA's prohibitory provisions should be reserved for combating intentional age-based discrimination.

A

Although Congress rejected proposals to address age discrimination in the Civil Rights Act of 1964, § 715 of that Act directed the Secretary of Labor to undertake a study of age *254discrimination in employment and to submit to Congress a report containing “such recommendations for legislation to prevent arbitrary discrimination in employment because of age as he determines advisable,” 78 Stat. 265. See General Dynamics Land Systems, Inc. v. Cline, 540 U. S. 581, 586-587 (2004); EEOC v. Wyoming, 460 U. S., at 229. In response, Secretary Willard Wirtz submitted the report that provided the blueprint for the ADEA. See Report of the Secretary of Labor, The Older American Worker: Age Discrimination in Employment (June 1965), reprinted in U. S. Equal Employment Opportunity Commission, Legislative History of the Age Discrimination in Employment Act (1981), Doc. No. 5 (hereinafter Wirtz Report or Report). Because the ADEA was modeled on the Wirtz Report’s findings and recommendations, the Report provides critical insights into the statute’s meaning. See generally Blumrosen, Interpreting the ADEA: Intent or Impact 14-20, in Age Discrimination in Employment Act: A Compliance and Litigation Manual for Lawyers and Personnel Practitioners 88-89 (M. Lake ed. 1982); see also General Dynamics, supra, at 587-590 (relying on the Wirtz Report to interpret the ADEA); EEOC v. Wyoming, supra, at 230-231 (discussing the Report’s role in the drafting of the ADEA).

The Wirtz Report reached two conclusions of central relevance to the question presented by this case. First, the Report emphasized that age discrimination is qualitatively different from the types of discrimination prohibited by Title VII of the Civil Rights Act of 1964 (i. e., race, color, religion, sex, and national origin discrimination). Most importantly — in stark contrast to the types of discrimination addressed by Title VII — the Report found no evidence that age discrimination resulted from intolerance or animus toward older workers. Rather, age discrimination was based primarily upon unfounded assumptions about the relationship between an individual’s age and her ability to perform a job. Wirtz Report 2. In addition, whereas ability is nearly al*255ways completely unrelated to the characteristics protected by Title VII, the Report found that, in some cases, “there is in fact a relationship between [an individual’s] age and his ability to perform the job.” Ibid, (emphasis deleted).

Second, the Wirtz Report drew a sharp distinction between “ ‘arbitrary discrimination’ ” (which the Report clearly equates with disparate treatment) and circumstances or practices having a disparate impact on older workers. See id., at 2, 21-22. The Report defined “arbitrary” discrimination as adverse treatment of older workers “because of assumptions about the effect of age on their ability to do a job when there is in fact no basis for these assumptions” Id., at 2 (emphasis in original). While the “most obvious kind” of arbitrary discrimination is the setting of unjustified maximum age limits for employment, id., at 6, naturally the Report’s definition encompasses a broad range of disparate treatment.

The Report distinguished such “arbitrary” (i. e., intentional and unfounded) discrimination from two other phenomena. One involves differentiation of employees based on a genuine relationship between age and ability to perform a job. See id., at 2. In this connection, the Report examined “circumstances which unquestionably affect older workers more strongly, as a group, than they do younger workers,” including questions of health, educational attainment, and technological change. Id., at 11-14.1 In addition, the Re*256port assessed “institutional arrangements” — such as seniority rules, workers’ compensation laws, and pension plans— which, though intended to benefit older workers, might actually make employers less likely to hire or retain them. Id., at 2, 15-17.

The Report specifically recommended legislative action to prohibit “arbitrary discrimination,” i. e., disparate treatment. Id., at 21-22. In sharp contrast, it recommended that the other two types of “discrimination” — both involving factors or practices having a disparate impact on older workers — be addressed through noncoercive measures: programs to increase the availability of employment; continuing education; and adjustment of pension systems, workers’ compensation, and other institutional arrangements. Id., at 22-25. These recommendations found direct expression in the ADEA, which was drafted at Congress’ command that the Secretary of Labor make “specific legislative recommendations for implementing the [Wirtz Report’s] conclusions,” Fair Labor Standards Amendments of 1966, § 606, 80 Stat. 845. See also General Dynamics, supra, at 589 (“[T]he ADEA . . . begins with statements of purpose and findings that mirror the Wirtz Report”).

B

The ADEA’s structure confirms Congress’ determination to prohibit only “arbitrary” discrimination (i. e., disparate treatment based on unfounded assumptions), while addressing practices with a disparate adverse impact on older work*257ers through noncoercive measures. Section 2 — which sets forth the findings and purposes of the statute — draws a clear distinction between “the setting of arbitrary age limits regardless of potential for job performance” and “certain otherwise desirable practices [that] may work to the disadvantage of older persons.” 29 U. S. C. § 621(a)(2). In response to these problems, § 2 identifies three purposes of the ADE A: “[1] to promote employment of older persons based on their ability rather than age; [2] to prohibit arbitrary age discrimination in employment; [and 3] to help employers and workers find ways of meeting problems arising from the impact of age on employment.” § 621(b).

Each of these three purposes corresponds to one of the three substantive statutory sections that follow. Section 3 seeks to “promote employment of older persons” by directing the Secretary of Labor to undertake a program of research and education related to “the needs and abilities of older workers, and their potentials for continued employment and contribution to the economy.” § 622(a). Section 4, which contains the ADE As core prohibitions, corresponds to the second purpose: to “prohibit arbitrary age discrimination in employment.” Finally, §5 addresses the third statutory purpose by requiring the Secretary of Labor to undertake a study of “institutional and other arrangements giving rise to involuntary retirement” and to submit any resulting findings and legislative recommendations to Congress. § 624(a)(1).

Section 4 — including § 4(a)(2) — must be read in light of the express statutory purpose the provision was intended to effect: the prohibition of “arbitrary age discrimination in employment.” § 621(b). As the legislative history makes plain, “arbitrary” age discrimination had a very specific meaning for the ADE A’s drafters. It meant disparate treatment of older workers, predominantly because of unfounded assumptions about the relationship between age and ability. See supra, at 255-256. Again, such intentional discrimination was clearly distinguished from circumstances and prac*258tices merely having a disparate impact on older workers, which — as ADEA §§2, 3, and 5 make clear — Congress intended to address through research, education, and possible future legislative action.

C

In addition to this affirmative evidence of congressional intent, I find it telling that the legislative history is devoid of any discussion of disparate impact claims or of the complicated issues such claims raise in the ADEA context. See Gold, Disparate Impact Under the Age Discrimination in Employment Act of 1967, 25 Berkeley J. Emp. & Lab. L. 1, 40 (2004). At the time the ADEA was enacted, the predominant focus of antidiscrimination law was on intentional discrimination; the concept of disparate impact liability, by contrast, was quite novel. See, e.g., Gold, Griggs’ Folly: An Essay on the Theory, Problems, and Origin of the Adverse Impact Definition of Employment Discrimination and a Recommendation for Reform, 7 Indus. Rel. L. J. 429, 518-520 (1985); Blumrosen, Strangers in Paradise: Griggs v. Duke Power Co. and the Concept of Employment Discrimination, 71 Mich. L. Rev. 59, 69-71 (1972-1973). Had Congress intended to inaugurate disparate impact liability in the ADEA, one would expect to find some indication of that intent in the text and the legislative history. There is none.

D

Congress’ decision not to authorize disparate impact claims is understandable in light of the questionable utility of such claims in the age-discrimination context. No one would argue that older workers have suffered disadvantages as a result of entrenched historical patterns of discrimination, like racial minorities have. See Massachusetts Bd. of Retirement v. Murgia, 427 U. S. 307, 313-314 (1976) (per curiam); see also Wirtz Report 5-6. Accordingly, disparate impact liability under the ADEA cannot be justified, and is not necessary, as a means of redressing the cumulative re-*259suits of past discrimination. Cf. Griggs, 401 U. S., at 430 (reasoning that disparate impact liability is necessary under Title VII to prevent perpetuation of the results of past racial discrimination).

Moreover, the Wirtz Report correctly concluded that — unlike the classifications protected by Title VII — there often is a correlation between an individual’s age and her ability to perform a job. Wirtz Report 2, 11-15. That is to be expected, for “physical ability generally declines with age,” Murgia, supra, at 315, and in some cases, so does mental capacity, see Gregory v. Ashcroft, 501 U. S. 452, 472 (1991). Perhaps more importantly, advances in technology and increasing access to formal education often leave older workers at a competitive disadvantage vis-a-vis younger workers. Wirtz Report 11-15. Beyond these performance-affecting factors, there is also the fact that many employment benefits, such as salary, vacation time, and so forth, increase as an employee gains experience and seniority. See, e.g., Finnegan v. Trans World Airlines, Inc., 967 F. 2d 1161, 1164 (CA7 1992) (“[Virtually all elements of a standard compensation package are positively correlated with age”). Accordingly, many employer decisions that are intended to cut costs or respond to market forces will likely have a disproportionate effect on older workers. Given the myriad ways in which legitimate business practices can have a disparate impact on older workers, it is hardly surprising that Congress declined to subject employers to civil liability based solely on such effects.

Ill

The plurality and Justice Scalia offer two principal arguments in favor of their reading of the statute: that the relevant provision of the ADE A should be read in pari mate-ria with the parallel provision of Title VII, and that we should give interpretive weight or deference to agency statements relating to disparate impact liability. I find neither argument persuasive.

*260A

The language of the ADEA’s prohibitory provisions was modeled on, and is nearly identical to, parallel provisions in Title VII. See McKennon v. Nashville Banner Publishing Co., 513 U. S. 352, 357 (1995); Lorillard v. Pons, 434 U. S. 575, 584 (1978). Because Griggs, supra, held that Title VII’s § 703(a)(2) permits disparate impact claims, the plurality concludes that we should read § 4(a)(2) of the ADEA similarly. Ante, at 233-238.

Obviously, this argument would be a great deal more convincing had Griggs been decided before the ADEA was enacted. In that case, we could safely assume that Congress had notice (and therefore intended) that the language at issue here would be read to authorize disparate impact claims. See, e. g., Department of Energy v. Ohio, 503 U. S. 607, 626 (1992); Holmes v. Securities Investor Protection Corporation, 503 U. S. 258, 268 (1992). But Griggs was decided four years after the ADEA’s enactment, and there is no reason to suppose that Congress in 1967 could have foreseen the interpretation of Title VII that was to come. See Fogerty v. Fantasy, Inc., 510 U. S. 517, 523, n. 9 (1994); see also supra, at 258 (discussing novelty of disparate impact theory at the time of the ADEA’s enactment).

To be sure, where two statutes use similar language we generally take this as “a strong indication that [they] should be interpreted pari passu.” Northcross v. Board of Ed. of Memphis City Schools, 412 U. S. 427, 428 (1973) (per curiam). But this is not a rigid or absolute rule, and it “‘readily yields’” to other indicia of congressional intent. General Dynamics, 540 U. S., at 595 (quoting Atlantic Cleaners & Dyers, Inc. v. United States, 286 U. S. 427, 433 (1932)). Indeed, “‘the meaning [of the same words] well may vary to meet the purposes of the law.’ ” United States v. Cleveland Indians Baseball Co., 532 U. S. 200, 213 (2001) (quoting Atlantic Cleaners & Dyers, supra, at 433; alteration in original). Accordingly, we have not hesitated to give *261a different reading to the same language — whether appearing in separate statutes or in separate provisions of the same statute — if there is strong evidence that Congress did not intend the language to be used uniformly. See, e. g., General Dynamics, supra, at 595-597 (“age” has different meaning where used in different parts of the ADEA); Cleveland Indians, supra, at 213 (“wages paid” has different meanings in different provisions of Title 26 U. S. C.); Robinson v. Shell Oil Co., 519 U. S. 337, 343-344 (1997) (“employee” has different meanings in different parts of Title VII); Fogerty, supra, at 522-525 (Copyright Act’s attorney’s fees provision has different meaning than the analogous provision in Title VII, despite their» “virtually identical language”). Such is the case here.

First, there are significant textual differences between Title VII and the ADEA that indicate differences in congressional intent. Most importantly, whereas the ADEA’s RFOA provision protects employers from liability for any actions not motivated by age, see supra, at 251-253, Title VII lacks any similar provision. In addition, the ADEA’s structure demonstrates Congress’ intent to combat intentional discrimination through §4’s prohibitions while addressing employment practices having a disparate impact on older workers through independent noncoercive mechanisms. See supra, at 256-258. There is no analogy in the structure of Title VII. Furthermore, as the Congresses that adopted both Title VII and the ADEA clearly recognized, the two statutes were intended to address qualitatively different kinds of discrimination. See supra, at 253-255. Disparate impact liability may have a legitimate role in combating the types of discrimination addressed by Title VII, but the nature of aging and of age discrimination makes such liability inappropriate for the ADEA. See supra, at 258-259.

Finally, nothing in the Court’s decision in Griggs itself provides any reason to extend its holding to the ADEA. As the plurality tacitly acknowledges, ante, at 235, the decision *262in Griggs was not based on any analysis of Title VIPs actual language. Rather, the ratio decidendi was the statute’s perceived purpose, i. e.,

“to achieve equality of employment opportunities and remove barriers that have operated in the past to favor an identifiable group of white employees over other employees. Under the Act, practices, procedures, or tests neutral on their face, and even neutral in terms of intent, cannot be maintained if they operate to ‘freeze’ the status quo of prior discriminatory employment practices.” 401 U. S., at 429-430.

In other words, the Court in Griggs reasoned that disparate impact liability was necessary to achieve Title VII’s ostensible goal of eliminating the cumulative effects of historical racial discrimination. However, that rationale finds no parallel in the ADEA context, see Murgia, 427 U. S., at 313-314, and it therefore should not control our decision here.

Even venerable canons of construction must bow, in an appropriate case, to compelling evidence of congressional intent. In my judgment, the significant differences between Title VII and the ADEA are more than sufficient to overcome the default presumption that similar language is to be read similarly. See Fogerty, supra, at 523-524 (concluding that the “normal indication” that similar language should be read similarly is “overborne” by differences between the legislative history and purposes of two statutes).

B

The plurality asserts that the agencies charged with the ADEA’s administration “have consistently interpreted the [statute] to authorize relief on a disparate-impact theory.” Ante, at 239. In support of this claim, the plurality describes a 1968 interpretive bulletin issued by the Department of Labor as “permitting]” disparate impact claims. Ibid, (citing 29 CFR § 860.103(f)(l)(i) (1970)). And the plu*263rality cites, without comment, an Equal Employment Opportunity Commission (EEOC) policy statement construing the RFOA provision. Ante, at 240 (citing 29 CFR §1625.7 (2004)). It is unclear what interpretive value the plurality means to assign to these agency statements. But Justice Scalia, at least, thinks that the EEOC statement is entitled to deference under Chevron U. S. A. Inc. v. Natural Resources Defense Council, Inc., 467 U. S. 837 (1984), and that “that is sufficient to resolve this case.” Ante, at 247 (opinion concurring in part and concurring in judgment). I disagree and, for the reasons that follow, would give no weight to the statements in question.

The 1968 Labor Department bulletin to which the plurality alludes was intended to “provide ‘a practical guide to employers and employees as to how the office representing the public interest in its enforcement will seek to apply it.’ ” 29 CFR §860.1 (1970) (quoting Skidmore v. Swift & Co., 323 U. S. 134, 138 (1944)). In discussing the RFOA provision, the bulletin states that “physical fitness requirements” and “[evaluation factors such as quantity or quality of production, or educational level” can qualify as reasonable nonage factors, so long as they have a valid relationship to job qualifications and are uniformly applied. §§ 860.103(f)(1), (2). But the bulletin does not construe the ADEA’s prohibitory provisions, nor does it state or imply that § 4(a) authorizes disparate impact claims. Rather, it establishes “a nonexclusive objective test for employers to use in determining whether they could be certain of qualifying for the” RFOA exemption. Public Employees Retirement System of Ohio v. Betts, 492 U. S. 158, 172 (1989) (discussing 1968 bulletin’s interpretation of the § 4(f)(2) exemption). Moreover, the very same bulletin states unequivocally that “[t]he clear purpose [of the ADEA] is to insure that age, within the limits prescribed by the Act, is not a determining factor in making any decision regarding the hiring, dismissal, promotion or any other term, condition or privilege of employment of an *264individual.” § 860.103(c) (emphasis added). That language is all about discriminatory intent.

The EEOC statement cited by the plurality and relied upon by Justice Scalia is equally unhelpful. This “interpretative rule or policy statement,” promulgated in 1981, superseded the 1968 Labor Department bulletin after responsibility for enforcing the ADEA was transferred from Labor to the EEOC. See 46 Fed. Reg. 47724 (1981). It states, in relevant part:

“[W]hen an employment practice, including a test, is claimed as a basis for different treatment of employees or applicants for employment on the grounds that it is a Tactor other than’ age, and such a practice has an adverse impact on individuals within the protected age group, it can only be justified as a business necessity.” 29 CFR § 1625.7(d) (2004).

Like the 1968 bulletin it replaces, this statement merely spells out the agency’s view, for purposes of its enforcement policy, of what an employer must do to be certain of gaining the safety of the RFOA haven. It says nothing about whether disparate impact claims are authorized by the ADEA.

For Justice Scalia, “[t]his is an absolutely classic case for deference to agency interpretation.” Ante, at 243 (opinion concurring in part and concurring in judgment). I disagree. Under Chevron, we will defer to a reasonable agency interpretation of ambiguous statutory language, see 467 U. S., at 843-844, provided that the interpretation has the requisite “force of law,” Christensen v. Harris County, 529 U. S. 576, 587 (2000). The rationale for such deference is that Congress has explicitly or implicitly delegated to the agency responsible for administering a statute the authority to choose among permissible constructions of ambiguous statutory text. See Chevron, supra, at 844. The question now before us is not what it takes to qualify for the RFOA exemption, *265but rather whether § 4(a)(2) of the ADEA authorizes disparate impact claims. But the EEOC statement does not purport to interpret the language of § 4(a) at all. Quite simply, the agency has not actually exercised its delegated authority to resolve any ambiguity in the relevant provision’s text, much less done so in a reasonable or persuasive manner. As to the specific question presented, therefore, the regulation is not entitled to any deference. See John Hancock Mut. Life Ins. Co. v. Harris Trust and Sav. Bank, 510 U. S. 86, 106-109, and n. 17 (1993); see also SEC v. Sloan, 436 U. S. 103, 117-118 (1978); Adamo Wrecking Co. v. United States, 434 U. S. 275, 287-289, and n. 5 (1978).2

Justice Scalia’s attempt to link the EEOC’s RFOA regulation to § 4(a)(2) is premised on a dubious chain of inferences that, in my view, highlights the hazards of his approach. Because the RFOA provision is “relevant only as a response to employer actions ‘otherwise prohibited’ by the ADEA,” he reasons, the “unavoidable meaning” of the EEOC statement is that the agency interprets the ADEA to prohibit “employer actions that have an ‘adverse impact on individuals within the protected age group.’” Ante, at 246 (opinion concurring in part and concurring in judgment) (quoting 29 CFR § 1625.7(d) (2004)). But, of course, disparate treatment clearly has an “adverse impact on individuals within the protected age group,” ibid., and Justice Scalia’s reading of the EEOC’s rule is hardly “unavoidable.” The regulation says only that if an employer wants to rely on a practice — say, a physical fitness test — as the basis for an exemption from liability, and that test adversely affects older workers, the employer can be sure of qualifying for the exemption only if the test is sufficiently job related. Such a *266limitation makes sense in disparate treatment cases. A test that harms older workers and is unrelated to the job may be a pretext for — or even a means of effectuating — intentional discrimination. See supra, at 253. Justice Scalia completes his analytical chain by inferring that the EEOC regulation must be read to interpret §U(a)(2) to allow disparate impact claims because that is the only provision of the ADEA that could “conceivably” be so interpreted. Ante, at 246. But the support for that inference is doubtful, to say the least. The regulation specifically refers to employment practices claimed as a basis for “different treatment of employees or applicants for employment,” 29 CFR § 1625.7(d) (2004) (emphasis added). Section 4(a)(2), of course, does not apply to “applicants for employment” at all — it is only § 4(a)(1) that protects this group. See 29 U. S. C. § 623(a). That suggests that the EEOC must have read the RFOA to provide a defense against claims under § 4(a)(1) — which unquestionably permits only disparate treatment claims, see supra, at 249.

This discussion serves to illustrate why it makes little sense to attribute to the agency a construction of the relevant statutory text that the agency itself has not actually articulated so that we can then “defer” to that reading. Such an approach is particularly troubling where applied to a question as weighty as whether a statute does or does not subject employers to liability absent discriminatory intent. This is not, in my view, what Chevron contemplated.

As an interpretation of the RFOA provision, moreover, the EEOC regulation is both unreasonable on its face and directly at odds with the Court’s holding in today’s case. It says that the RFOA exemption is available only if the employer’s practice is justified by a “business necessity.” But the Court has rejected that reading of the RFOA provision, and rightly so: There may be many “reasonable” means by which an employer can advance its goals, and a given nonage factor can certainly be “reasonable” without being necessary. *267 Ante, at 243; see also Western Air Lines, Inc. v. Criswell, 472 U. S. 400, 419 (1985) (distinguishing “ ‘reasonable necessity’ ” standard from “reasonableness”). Of course, it is elementary that “no deference is due to agency interpretations at odds with the plain language of the statute itself.” Betts, 492 U. S., at 171. The agency clearly misread the RFOA provision it was attempting to construe. That error is not necessarily dispositive of the disparate impact question. But I think it highlights the improvidence of giving weight (let alone deferring) to the regulation’s purported assumption that an entirely different provision of the statute, which is not even the subject of the regulation, authorizes disparate impact claims. In my view, we should simply acknowledge that this regulation is of no help in answering the question presented.

IV

Although I would not read the ADEA to authorize disparate impact claims, I agree with the Court that, if such claims are allowed, they are strictly circumscribed by the RFOA exemption. See ante, at 241-242. That exemption requires only that the challenged employment practice be based on a “reasonable” nonage factor — that is, one that is rationally related to some legitimate business objective. I also agree with the Court, ante, at 240, that, if disparate impact claims are to be permitted under the ADEA, they are governed by the standards set forth in our decision in Wards Cove Packing Co. v. Atonio, 490 U. S. 642 (1989). That means, as the Court holds, ante, at 241, that “a plaintiff must demonstrate that it is the application of a specific or particular employment practice that has created the disparate impact under attack,” Wards Cove, supra, at 657 (emphasis added); see also Watson v. Fort Worth Bank & Trust, 487 U. S. 977, 994 (1988) (opinion of O’Connor, J.). It also means that once the employer has produced evidence that its action was based on a reasonable nonage factor, the plaintiff bears the burden of disproving this assertion. See Wards Cove, supra, at 659-*268660; see also Watson, supra, at 997 (opinion of O’Connor, J.). Even if petitioners’ disparate impact claim were cognizable under the ADEA, that claim clearly would fail in light of these requirements.