concurring.
I join the Court’s opinion in full, but write separately to observe that its resolution of this dispute merely postpones the evil day on which the Court will have to confront the question: Whether, or to what extent, are the disparate-impact provisions of Title VII of the Civil Rights Act of 1964 consistent with the Constitution’s guarantee of equal protection? The question is not an easy one. See generally Primus, Equal Protection and Disparate Impact: Round Three, 117 Harv. L. Rev. 493 (2003).
The difficulty is this: Whether or not Title VII’s disparate-treatment provisions forbid “remedial” race-based actions when a disparate-impact violation would not otherwise result — the question resolved by the Court today — it is clear that Title VII not only permits but affirmatively requires such actions when a disparate-impact violation would otherwise result. See ante, at 580-581. But if the Federal Government is prohibited from discriminating on the basis of race, Bolling v. Sharpe, 347 U. S. 497, 500 (1954), then surely it is also prohibited from enacting laws mandating that third parties — e. g., employers, whether private, state, or municipal — discriminate on the basis of race. See Buchanan v. Warley, 245 U. S. 60, 78-82 (1917). As the facts of these cases illustrate, Title VII’s disparate-impact provisions place a racial thumb on the scales, often requiring employers to evaluate the racial outcomes of their policies, and to make decisions based on (because of) those racial outcomes. That type of racial decisionmaking is, as the Court explains, discriminatory. See ante, at 578-579; Personnel Administrator of Mass. v. Feeney, 442 U. S. 256, 279 (1979).
To be sure, the disparate-impact laws do not mandate imposition of quotas, but it is not clear why that should provide a safe harbor. Would a private employer not be guilty of unlawful discrimination if he refrained from establishing a racial hiring quota but intentionally designed his hiring practices to achieve the same end? Surely he would. In*595tentional discrimination is still occurring, just one step up the chain. Government compulsion of such design would therefore seemingly violate equal protection principles. Nor would it matter that Title VII requires consideration of race on a wholesale, rather than retail, level. “[T]he Government must treat citizens as individuals, not as simply components of a racial, religious, sexual or national class.” Miller v. Johnson, 515 U. S. 900, 911 (1995) (internal quotation marks omitted). And of course the purportedly benign motive for the disparate-impact provisions cannot save the statute. See Adarand Constructors, Inc. v. Peña, 515 U. S. 200, 227 (1995).
It might be possible to defend the law by framing it as simply an evidentiary tool used to identify genuine, intentional discrimination — to “smoke out,” as it were, disparate treatment. See Primus, supra, at 498-499, 520-521. Disparate impact is sometimes (though not always, see Watson v. Fort Worth Bank & Trust, 487 U. S. 977, 992 (1988) (plurality opinion)) a signal of something illicit, so a regulator might allow statistical disparities to play some role in the evidentiary process. Cf. McDonnell Douglas Corp. v. Green, 411 U. S. 792, 802-803 (1973). But arguably the disparate-impact provisions sweep too broadly to be fairly characterized in such a fashion — since they fail to provide an affirmative defense for good-faith (i. e., nonracially motivated) conduct, or perhaps even for good faith plus hiring standards that are entirely reasonable. See post, at 621-623, and n. 3 (Ginsburg, J., dissenting) (describing the demanding nature of the “business necessity” defense). This is a question that this Court will have to consider in due course. It is one thing to free plaintiffs from proving an employer’s illicit intent, but quite another to preclude the employer from proving that its motives were pure and its actions reasonable.
The Court’s resolution of these cases makes it unnecessary to resolve these matters today. But the war between disparate impact and equal protection will be waged sooner or *596later, and it behooves us to begin thinking about how — and on what terms — to make peace between them.
Justice Alito, with whom Justice Scalia and Justice Thomas join, concurring.
I join the Court’s opinion in full. I write separately only because the dissent, while claiming that “[t]he Court’s recitation of the facts leaves out important parts of the story,” post, at 609 (opinion of Ginsburg, J.), provides an incomplete description of the events that led to New Haven’s decision to reject the results of its exam. The dissent’s omissions are important because, when all of the evidence in the record is taken into account, it is clear that, even if the legal analysis in Parts II and III-A of the dissent were accepted, affirmance of the decision below is untenable.
I
When an employer in a disparate-treatment case under Title VII of the Civil Rights Act of 1964 claims that an employment decision, such as the refusal to promote, was based on a legitimate reason, two questions — one objective and one subjective — must be decided. The first, objective question is whether the reason given by the employer is one that is legitimate under Title VII. See St. Mary’s Honor Center v. Hicks, 509 U. S. 502, 506-507 (1993). If the reason provided by the employer is not legitimate on its face, the employer is liable. Id., at 509. The second, subjective question concerns the employer’s intent. If an employer offers a facially legitimate reason for its decision but it turns out that this explanation was just a pretext for discrimination, the employer is again liable. See id., at 510-512.
The question on which the opinion of the Court and the dissenting opinion disagree concerns the objective component of the determination that must be made when an employer justifies an employment decision, like the one made in *597this litigation, on the ground that a contrary decision would have created a risk of disparate-impact liability. The Court holds — and I entirely agree — that concern about disparate-impact liability is a legitimate reason for a decision of the type involved here only if there was a “strong basis in evidence to find the tests inadequate.” Ante, at 585. The Court ably demonstrates that in this litigation no reasonable jury could find that the city of New Haven (City) possessed such evidence and therefore summary judgment for petitioners is required. Because the Court correctly holds that respondents cannot satisfy this objective component, the Court has no need to discuss the question of respondents’ actual intent. As the Court puts it, “[e]ven if respondents were motivated as a subjective matter by a desire to avoid committing disparate-impact discrimination, the record makes clear there is no support for the conclusion that respondents had an objective, strong basis in evidence to find the tests inadequate.” Ibid.
The dissent advocates a different objective component of the governing standard. According to the dissent, the objective component should be whether the evidence provided “good cause” for the decision, post, at 625, and the dissent argues — incorrectly, in my view — that no reasonable juror could fail to find that such evidence was present here. But even if the dissent were correct on this point, I assume that the dissent would not countenance summary judgment for respondents if respondents’ professed concern about disparate-impact litigation was simply a pretext. Therefore, the decision below, which sustained the entry of summary judgment for respondents, cannot be affirmed unless no reasonable jury could find that the City’s asserted reason for scrapping its test — concern about disparate-impact liability — was a pretext and that the City’s real reason was illegitimate, namely, the desire to placate a politically important racial constituency.
*598II
A
As initially described by the dissent, see post, at 609-618, the process by which the City reached the decision not to accept the test results was open, honest, serious, and deliberative. But even the District Court admitted that “a jury could rationally infer that city officials worked behind the scenes to sabotage the promotional examinations because they knew that, were the exams certified, the Mayor would incur the wrath of [Rev. Boise] Kimber and other influential leaders of New Haven’s African-American community.” 554 F. Supp. 2d 142, 162 (Conn. 2006) (internal quotation marks omitted), summarily aff’d, 530 F. 3d 87 (CA2 2008) (per curiam).
This admission finds ample support in the record. Rev. Boise Kimber, to whom the District Court referred, is a politically powerful New Haven pastor and a self-professed “‘kingmaker.’” App. to Pet. for Cert, in No. 07-1428, p. 906a; see also id., at 909a. On one occasion, “[i]n front of TV cameras, he threatened a race riot during the murder trial of the black man arrested for killing white Yalie Christian Prince. He continues to call whites racist if they question his actions.” Id., at 931a.
Reverend Kimber’s personal ties with seven-term New Haven Mayor John DeStefano (Mayor) stretch back more than a decade. In 1996, for example, Mayor DeStefano testified for Reverend Kimber as a character witness when Reverend Kimber — then the manager of a funeral home — was prosecuted and convicted for stealing prepaid funeral expenses from an elderly woman and then lying about the matter under oath. See id., at 126a, 907a. “Reverend Kimber has played a leadership role in all of Mayor DeStefano’s political campaigns, [and] is considered a valuable political supporter and vote-getter.” Id., at 126a. According to the Mayor’s former campaign manager (who is currently his executive assistant), Reverend Kimber is an invaluable political *599asset because “[h]e’s very good at organizing people and putting together field operations, as a result of his ties to labor, his prominence in the religious community and his longstanding commitment to roots.” Id., at 908a (internal quotation marks and alteration omitted).
In 2002, the Mayor picked Reverend Kimber to serve as the chairman of the New Haven Board of Fire Commissioners (BFC), “despite the fact that he had no experience in the profession, fire administration, [or] municipal management.” Id., at 127a; see also id., at 928a-929a. In that capacity, Reverend Kimber told firefighters that certain new recruits would not be hired because “ ‘they just have too many vowels in their name[s].’” Thanawala, New Haven Fire Panel Chairman Steps Down Over Racial Slur, Hartford Courant, June 13, 2002, p. B2. After protests about this comment, Reverend Kimber stepped down as chairman of the BFC, ibid.; see also App. to Pet. for Cert, in No. 07-1428, at 929a, but he remained on the BFC and retained “a direct line to the mayor,” id., at 816a.
Almost immediately after the test results were revealed in “early January” 2004, Reverend Kimber called the City’s chief administrative officer, Karen Dubois-Walton, who “acts ‘on behalf of the Mayor.’” Id., at 221a, 812a. DuboisWalton and Reverend Kimber met privately in her office because he wanted “to express his opinion” about the test results and “to have some influence” over the City’s response. Id., at 815a-816a. As discussed in further detail below, Reverend Kimber adamantly opposed certification of the test results — a fact that he or someone in the Mayor’s office eventually conveyed to the Mayor. Id., at 229a.
B
On January 12, 2004, Tina Burgett (the director of the City’s Department of Human Resources) sent an e-mail to Dubois-Walton to coordinate the City’s response to the test results. Burgett wanted to clarify that the City’s executive *600officials would meet “sans the Chief, and that once we had a better fix on the next steps we would meet with the Mayor (possibly) and then the two Chiefs.” Id., at 446a. The “two Chiefs” are Fire Chief William Grant (who is white) and Assistant Fire Chief Ronald Dumas (who is African-American). Both chiefs believed that the test results should be certified. Id., at 228a, 817a. Petitioners allege, and the record suggests, that the Mayor and his staff colluded “sans the Chief[s]” because “the defendants did not want Grant’s and Dumas’ views to be expressed or known; accordingly both men were prevented by the Mayor and his staff from making any statements regarding the matter.” Id., at 228a.1
The next day, on January 13, 2004, Chad Legel, who had designed the tests, flew from Chicago to New Haven to meet with Dubois-Walton, Burgett, and Thomas Ude, the City’s corporate counsel. Id., at 179a. “Legel outlined the merits of the examination and why city officials should be confident in the validity of the results.” Ibid. But according to Legel, Dubois-Walton was “argumentative” and apparently had already made up her mind that the tests were “ ‘discriminatory.’ ” Id., at 179a-180a. Again according to Legel, “[a] theme” of the meeting was “the political and racial overtones of what was going on in the City.” Id., at 181a. “Legel came away from the January 13, 2004 meeting with the impression that defendants were already leaning toward discarding the examination results.” Id., at 180a.
On January 22, 2004, the Civil Service Board (CSB or Board) convened its first public meeting. Almost immediately, Reverend Kimber began to exert political pressure on the CSB. He began a loud, minutes-long outburst that required the CSB chairman to shout him down and hold him out of order three times. See id., at 187a, 467a-468a; see *601also App. in No. 06-4996-cv (CA2), pp. A703-A705. Reverend Kimber protested the public meeting, arguing that he and the other fire commissioners should first be allowed to meet with the CSB in private. App. to Pet. for Cert. in No. 07-1428, at 188a.
Four days after the CSB’s first meeting, Mayor DeStefano’s executive aide sent an e-mail to Dubois-Walton, Burgett, and Ude. Id., at 190a. The message clearly indicated that the Mayor had made up his mind to oppose certification of the test results (but nevertheless wanted to conceal that fact from the public):
“I wanted to make sure we are all on the same page for this meeting tomorrow. . . . [LJet’s remember, that these folks are not against certification yet. So we can’t go in and tell them that is our position; we have to deliberate and arrive there as the fairest and most cogent outcome.” Ibid.
On February 5, 2004, the CSB convened its second public meeting. Reverend Kimber again testified and threatened the CSB with political recriminations if they voted to certify the test results:
“I look at this [Board] tonight. I look at three whites and one Hispanic and no blacks. ... I would hope that you would not put yourself in this type of position, a political ramification that may come back upon you as you sit on this [Board] and decide the future of a department and the future of those who are being promoted.
“(APPLAUSE).” Id., at 492a (emphasis added).
One of the CSB members “t[ook] great offense” because he believed that Reverend Kimber “considered] [him] a bigot because [his] face is white.” Id., at 496a. The offended *602CSB member eventually voted not to certify the test results. Id., at 586a-587a.
One of Reverend Kimber’s “friends and allies,” Lieutenant Gary Tinney, also exacerbated racial tensions before the CSB. Id., at 129a. After some firefighters applauded in support of certifying the test results, “Lt. Tinney exclaimed, ‘Listen to the Klansmen behind us.’” Id., at 225a.
Tinney also has strong ties to the Mayor’s office. See, e. g., id., at 129a-130a, 816a-817a. After learning that he had not scored well enough on the captain’s exam to earn a promotion, Tinney called Dubois-Walton and arranged a meeting in her office. Id., at 830a-831a, 836a. Tinney alleged that the white firefighters had cheated on their exams — an accusation that Dubois-Walton conveyed to the Board without first conducting an investigation into its veracity. Id., at 837a-838a; see also App. 164 (statement of CSB chairman, noting the allegations of cheating). The allegation turned out to be baseless. App. to Pet. for Cert. in No. 07-1428, at 836a.
Dubois-Walton never retracted the cheating allegation, but she and other executive officials testified several times before the CSB. In accordance with directions from the May- or’s office to make the CSB meetings appear deliberative, see id., at 190a, executive officials remained publicly uncommitted about certification — while simultaneously “work[ing] as a team” behind closed doors with the secretary of the CSB to devise a political message that would convince the CSB to vote against certification, see id., at 447a. At the public CSB meeting on March 11, 2004, for example, Corporation Counsel Ude bristled at one board member’s suggestion that City officials were recommending against certifying the test results. See id., at 215a (“Attorney Ude took offense, stating, ‘Frankly, because I would never make a recommendation — I would not have made a recommendation like that’ ”). But within days of making that public statement, Ude privately told other members of the Mayor’s team “the ONLY *603way we get to a decision not to certify is” to focus on something other than “a big discussion re: adverse impact” law. Id., at 458a-459a.
As part of its effort to deflect attention from the specifics of the test, the City relied heavily on the testimony of Dr. Christopher Hornick, who is one of Chad Legel’s competitors in the test-development business. Hornick never “stud-tied] the test [that Legel developed] at length or in detail,” id., at 549a; see also id., at 203a, 553a, but Hornick did review and rely upon literature sent to him by Burgett to criticize Legel’s test. For example, Hornick “noted in the literature that [Burgett] sent that the test was not customized to the New Haven Fire Department.” Id., at 551a. The chairman of the CSB immediately corrected Hornick. Id., at 552a (“Actually, it was, Dr. Hornick”). Hornick also relied on newspaper accounts — again, sent to him by Burgett — pertaining to the controversy surrounding the certification decision. See id., at 204a, 557a. Although Hornick again admitted that he had no knowledge about the actual test that Legel had developed and that the City had administered, see id., at 560a-561a, the City repeatedly relied upon Hornick as a testing “guru” and, in the CSB chairman’s words, “the City ke[pt] quoting him as a person that we should rely upon more than anybody else [to conclude that there] is a better way— a better mousetrap.”2 App. in No. 06-4996-cv (CA2), at A1128. Dubois-Walton later admitted that the City rewarded Hornick for his testimony by hiring him to develop and administer an alternative test. App. to Pet. for Cert. in *604No. 07-1428, at 854a; see also id., at 562a-563a (Hornick’s plea for future business from the City on the basis of his criticisms of Legel’s tests).
At some point prior to the CSB’s public meeting on March 18,2004, the Mayor decided to use his executive authority to disregard the test results — even if the CSB ultimately voted to certify them. Id., at 819a-820a. Accordingly, on the evening of March 17th, Dubois-Walton sent an e-mail to the Mayor, the Mayor’s executive assistant, Burgett, and attorney Ude, attaching two alternative press releases. Id., at 457a. The first would be issued if the CSB voted not to certify the test results; the second would be issued (and would explain the Mayor’s invocation of his executive authority) if the CSB voted to certify the test results. Id., at 217a-218a, 590a-591a, 819a-820a. Half an hour after Dubois-Walton circulated the alternative drafts, Burgett replied: “[W]ell, that seems to say it all. Let’s hope draft #2 hits the shredder tomorrow nite.” Id., at 457a.
Soon after the CSB voted against certification, Mayor De-Stefano appeared at a dinner event and “took credit for the scu[tt]ling of the examination results.” Id., at 230a.
C
Taking into account all the evidence in the summary judgment record, a reasonable jury could find the following. Almost as soon as the City disclosed the racial makeup of the list of firefighters who scored the highest on the exam, the City administration was lobbied by an influential community leader to scrap the test results, and the City administration decided on that course of action before making any real assessment of the possibility of a disparate-impact violation. To achieve that end, the City administration concealed its internal decision but worked — as things turned out, successfully — to persuade the CSB that acceptance of the test results would be illegal and would expose the City to disparate-impact liability. But in the event that the CSB *605was not persuaded, the Mayor, wielding ultimate decision-making authority, was prepared to overrule the CSB immediately. Taking this view of the evidence, a reasonable jury could easily find that the City’s real reason for scrapping the test results was not a concern about violating the disparate-impact provision of Title VII but a simple desire to please a politically important racial constituency. It is noteworthy that the Solicitor General — whose position on the principal legal issue here is largely aligned with the dissent — eon-' eludes that “[n]either the district court nor the court of appeals . . . adequately considered whether, viewing the evidence in the light most favorable to petitioners, a genuine issue of material fact remained whether respondents’ claimed purpose to comply with Title VII was a pretext for intentional racial discrimination . . . .” Brief for United States as Amicus Curiae 6; see also id., at 32-33.
Ill
I will not comment at length on the dissent’s criticism of my analysis, but two points require a response.
The first concerns the dissent’s statement that I “equat[e] political considerations with unlawful discrimination.” Post, at 642. The dissent misrepresents my position: I draw no such equation. Of course “there are many ways in which a politician can attempt to win over a constituency — including a racial constituency — without engaging in unlawful discrimination.” Ibid. But — as I assume the dissent would agree — there are some things that a public official cannot do, and one of those is engaging in intentional racial discrimination when making employment decisions.
The second point concerns the dissent’s main argument— that efforts by the Mayor and his staff to scuttle the test results are irrelevant because the ultimate decision was made by the CSB. According to the dissent, “[t]he relevant decision was made by the CSB,” post, at 640, and there is “scant cause to suspect” that anything done by the opponents *606of certification, including the Mayor and his staff, “prevented the CSB from evenhandedly assessing the reliability of the exams and rendering an independent, good-faith decision on certification,” post, at 641.
Adoption of the dissent’s argument would implicitly decide an important question of Title VII law that this Court has never resolved — the circumstances in which an employer may be held liable based on the discriminatory intent of subordinate employees who influence but do not make the ultimate employment decision. There is a large body of Court of Appeals case law on this issue, and these cases disagree about the proper standard. See EEOC v. BCI Coca-Cola Bottling Co. of Los Angeles, 450 F. 3d 476, 484-488 (CA10 2006) (citing cases and describing the approaches taken in different Circuits). One standard is whether the subordinate “exerted influencie] over the titular decisionmaker.” Russell v. McKinney Hosp. Venture, 235 F. 3d 219, 227 (CA5 2000); see also Poland v. Chertoff, 494 F. 3d 1174, 1182 (CA9 2007) (A subordinate’s bias is imputed to the employer where the subordinate “influenced or was involved in the decision or decisionmaking process”). Another is whether the discriminatory input “caused the adverse employment action.” See BCI Coca-Cola Bottling Co. of Los Angeles, supra, at 487.
In the present cases, a reasonable jury could certainly find that these standards were met. The dissent makes much of the fact that members of the CSB swore under oath that their votes were based on the good-faith belief that certification of the results would have violated federal law. See post, at 640. But the good faith of the CSB members would not preclude a finding that the presentations engineered by the Mayor and his staff influenced or caused the CSB decision.
The least employee-friendly standard asks only whether “the actual decisionmaker” acted with discriminatory intent, see Hill v. Lockheed Martin Logistics Management, Inc., *607354 F. 3d 277, 291 (CA4 2004) (en banc), and it is telling that, even under this standard, summary judgment for respondents would not be proper. This is so because a reasonable jury could certainly find that in New Haven, the Mayor — not the CSB — wielded the final decisionmaking power. After all, the Mayor claimed that authority and was poised to use it in the event that the CSB decided to accept the test results. See supra, at 604. If the Mayor had the authority to overrule a CSB decision accepting the test results, the Mayor also presumably had the authority to overrule the CSB’s decision rejecting the test results. In light of the Mayor’s conduct, it would be quite wrong to throw out petitioners’ case on the ground that the CSB was the ultimate decisionmaker.
* * *
Petitioners are firefighters who seek only a fair chance to move up the ranks in their chosen profession. In order to qualify for promotion, they made personal sacrifices. Petitioner Frank Ricci, who is dyslexic, found it necessary to “hir[e] someone, at considerable expense, to read onto audiotape the content of the books and study materials].” App. to Pet. for Cert. in No. 07-1428, at 169a. He “studied an average of eight to thirteen hours a day ... , even listening to audio tapes while driving his car.” Ibid. Petitioner Benjamin Vargas, who is Hispanic, had to “give up a part-time job,” and his wife had to “take leave from her own job in order to take care of their three young children while Vargas studied.” Id., at 176a. “Vargas devoted countless hours to study..., missed two of his children’s birthdays and over two weeks of vacation time,” and “incurred significant financial expense” during the 3-month study period. Id., at 176a-177a.
Petitioners were denied promotions for which they qualified because of the race and ethnicity of the firefighters who achieved the highest scores on the City’s exam. The District Court threw out their case on summary judgment, even *608though that court all but conceded that a jury could find that the City’s asserted justification was pretextual. The Court of Appeals then summarily affirmed that decision.
The dissent grants that petitioners’ situation is “unfortunate” and that they “understandably attract this Court’s sympathy.” Post this page and 644. But “sympathy” is not what petitioners have a right to demand. What they have a right to demand is evenhanded enforcement of the law — of Title VII’s prohibition against discrimination based on race. And that is what, until today’s decision, has been denied them.
Justice Ginsburg, with whom Justice Stevens, Justice Souter, and Justice Breyer join, dissenting.
In assessing claims of race discrimination, “[c]ontext matters.” Grutter v. Bollinger, 539 U. S. 306, 327 (2003). In 1972, Congress extended Title VII of the Civil Rights Act of 1964 to cover public employment. At that time, municipal fire departments across the country, including New Haven’s, pervasively discriminated against minorities. The extension of Title VII to cover jobs in firefighting effected no overnight change. It took decades of persistent effort, advanced by Title VII litigation, to open firefighting posts to members of racial minorities.
The white firefighters who scored high on New Haven’s promotional exams understandably attract this Court’s sympathy. But they had no vested right to .promotion. Nor have other persons received promotions in preference to them. New Haven maintains that it refused to certify the test results because it believed, for good cause, that it would be vulnerable to a Title VII disparate-impact suit if it relied on those results. The Court today holds that New Haven has not demonstrated “a strong basis in evidence” for its plea. Ante, at 563. In so holding, the Court pretends that “[t]he City rejected the test results solely because the higher scoring candidates were white. ” Ante, at 580. That preten*609sion, essential to the Court’s disposition, ignores substantial evidence of multiple flaws in the tests New Haven used. The Court similarly fails to acknowledge the better tests used in other cities, which have yielded less racially skewed outcomes.1
By order of this Court, New Haven, a city in which African-Americans and Hispanics account for nearly 60 percent of the population, must today be served — as it was in the days of undisguised discrimination — by a fire department in which members of racial and ethnic minorities are rarely seen in command positions. In arriving at its order, the Court barely acknowledges the pathmarking decision in Griggs v. Duke Power Co., 401 U. S. 424 (1971), which explained the centrality of the disparate-impact concept to effective enforcement of Title VII. The Court’s order and opinion, I anticipate, will not have staying power.
I
A
The Court’s recitation of the facts leaves out important parts of the story. Firefighting is a profession in which the legacy of racial discrimination casts an especially long shadow. In extending Title VII to state and local government employers in 1972, Congress took note of a U. S. Commission on Civil Rights (USCCR) report finding racial discrimination in municipal employment even “more pervasive than in the private sector.” H. R. Rep. No. 92-288, p. 17 (1971). According to the report, overt racism was partly to blame, but so too was a failure on the part of municipal em*610ployers to apply merit-based employment principles. In making hiring and promotion decisions, public employers often “rel[ied] on criteria unrelated to job performance,” including nepotism or political patronage. 118 Cong. Rec. 1817 (1972). Such flawed selection methods served to entrench preexisting racial hierarchies. The USCCR report singled out police and fire departments for having “[bjarriers to equal employment. . . greater . . . than in any other area of State or local government,” with African-Americans “holding] almost no positions in the officer ranks.” Ibid. See also National Commission on Fire Prevention and Control, America Burning 5 (1973) (“Racial minorities are under-represented in the fire departments in nearly every community in which they live.”).
The city of New Haven (City) was no exception. In the early 1970’s, African-Americans and Hispanics composed 30 percent of New Haven’s population, but only 3.6 percent of the City’s 502 firefighters. The racial disparity in the officer ranks was even more pronounced: “[O]f the 107 officers in the Department only one was black, and he held the lowest rank above private.” Firebird Soc. of New Haven, Inc. v. New Haven Bd. of Fire Comm’rs, 66 F. R. D. 457, 460 (Conn. 1975).
Following a lawsuit and settlement agreement, see ibid., the City initiated efforts to increase minority representation in the New Haven Fire Department (Department). Those litigation-induced efforts produced some positive change. New Haven’s population includes a greater proportion of minorities today than it did in the 1970’s: Nearly 40 percent of the City’s residents are African-American and more than 20 percent are Hispanic. Among entry-level firefighters, minorities are still underrepresented, but not starkly so. As of 2003, African-Americans and Hispanics constituted 30 percent and 16 percent of the City’s firefighters, respectively. In supervisory positions, however, significant disparities remain. Overall, the senior officer ranks (captain and higher) *611are nine percent African-American and nine percent Hispanic. Only one of the Department’s 21 fire captains is African-American. See App. in No. 06-4996-cv (CA2), p. A1588 (hereinafter CA2 App.). It is against this backdrop of entrenched inequality that the promotion process at issue in this litigation should be assessed.
B
By order of its charter, New Haven must use competitive examinations to fill vacancies in fire-officer and other civil-service positions. Such examinations, the City’s civil-service rules specify, “shall be practical in nature, shall relate to matters which fairly measure the relative fitness and capacity of the applicants to discharge the duties of the position which they seek, and shall take into account character, training, experience, physical and mental fitness.” Id., at A331. The City may choose among a variety of testing methods, including written and oral exams and “[performance tests to demonstrate skill and ability in performing actual work.” Id., at A332.
New Haven, the record indicates, did not closely consider what sort of “practical” examination would “fairly measure the relative fitness and capacity of the applicants to discharge the duties” of a fire officer. Instead, the City simply adhered to the testing regime outlined in its two-decades-old contract with the local firefighters’ union: a written exam, which would account for 60 percent of an applicant’s total score, and an oral exam, which would account for the remaining 40 percent. Id., at A1045. In soliciting bids from exam development companies, New Haven made clear that it would entertain only “proposals that include a written component that will be weighted at 60%, and an oral component that will be weighted at 40%. ” Id., at A342. Chad Legel, a representative of the winning bidder, Industrial/Organizational Solutions, Inc. (IOS), testified during his deposition that the City never asked whether alternative methods *612might better measure the qualities of a successful fire officer, including leadership skills and command presence. See id., at A522 (“I was under contract and had responsibility only to create the oral interview and the written exam.”).
Pursuant to New Haven’s specifications, IOS developed and administered the oral and written exams. The results showed significant racial disparities. On the lieutenant exam, the pass rate for African-American candidates was about one-half the rate for Caucasian candidates; the pass rate for Hispanic candidates was even lower. On the captain exam, both African-American and Hispanic candidates passed at about half the rate of their Caucasian counterparts. See App. 225-226. More striking still, although nearly half of the 77 lieutenant candidates were African-American or Hispanic, none would have been eligible for promotion to the eight positions then vacant. The highest scoring African-American candidate ranked 13th; the top Hispanic candidate was 26th. As for the seven then-vacant captain positions, two Hispanic candidates would have been eligible, but no African-Americans. The highest scoring African-American candidate ranked 15th. See id., at 218-219.
These stark disparities, the Court acknowledges, sufficed to state a prima facie case under Title VII’s disparate-impact provision. See ante, at 586 (“The pass rates of minorities ... f[e]ll well below the 80-percent standard set by the [Equal Employment Opportunity Commission (EEOC)] to implement the disparate-impact provision of Title VII.”). New Haven thus had cause for concern about the prospect of Title VII litigation and liability. City officials referred the matter to the New Haven Civil Service Board (CSB), the entity responsible for certifying the results of employment exams.
Between January and March 2004, the CSB held five public meetings to consider the proper course. At the first meeting, New Haven’s Corporation Counsel, Thomas Ude, described the legal standard governing Title VII disparate-impact claims. Statistical imbalances alone, Ude correctly *613recognized, do not give rise to liability. Instead, presented with a disparity, an employer “has the opportunity and the burden of proving that the test is job-related and consistent with business necessity.” CA2 App. A724. A Title VII plaintiff may attempt to rebut an employer’s showing of job-relatedness and necessity by identifying alternative selection methods that would have been at least as valid but with “less of an adverse or disparate or discriminatory effect.” Ibid. See also id., at A738. Accordingly, the CSB commissioners understood, their principal task was to decide whether they were confident about the reliability of the exams: Had the exams fairly measured the qualities of a successful fire officer despite their disparate results? Might an alternative examination process have identified the most qualified candidates without creating sueh significant racial imbalances?
Seeking a range of input on these questions, the CSB heard from test takers, the test designer, subject-matter experts, City officials, union leaders, and community members. Several candidates for promotion, who did not yet know their exam results, spoke at the CSB’s first two meetings. Some candidates favored certification. The exams, they emphasized, had closely tracked the assigned study materials. Having invested substantial time and money to prepare themselves for the test, they, felt it would be unfair to scrap the results. See, e. g., id., at A772-A773, A785-A789.
Other firefighters had a different view. A number of the exam questions, they pointed out, were not germane to New Haven’s practices and procedures. See, e. g., id., at A774-A784. At least two candidates opposed to certification noted unequal access to study materials. Some individuals, they asserted, had the necessary books even before the syllabus was issued. Others had to invest substantial sums to purchase the materials and “wait a month and a half for some of the books because they were on back-order.” Id., at A858. These disparities, it was suggested, fell at least in part along racial lines. While many Caucasian applicants could obtain *614materials and assistance from relatives in the fire service, the overwhelming majority of minority applicants were “first-generation firefighters” without such support networks. See id., at A857-A861, A886-A887.
A representative of the Northeast Region of the International Association of Black Professional Firefighters, Donald Day, also spoke at the second meeting. Statistical disparities, he told the CSB, had been present in the Department’s previous promotional exams. On earlier tests, however, a few minority candidates had fared well enough to earn promotions. Id., at A828. See also App. 218-219. Day contrasted New Haven’s experience with that of nearby Bridgeport, where minority firefighters held one-third of lieutenant and captain positions. Bridgeport, Day observed, had once used a testing process similar to New Haven’s, with a written exam accounting for 70 percent of an applicant’s score, an oral exam for 25 percent, and seniority for the remaining five percent. CA2 App. A830. Bridgeport recognized, however, that the oral component, more so than the written component, addressed the sort of “real-life scenarios” fire officers encounter on the job. Id., at A832. Accordingly, that city “changed the relative weights” to give primacy to the oral exam. Ibid. Since that time, Day reported, Bridgeport had seen minorities “fairly represented” in its exam results. Ibid.
The CSB’s third meeting featured IOS representative Legel, the leader of the team that had designed and administered the exams for New Haven. Several City officials also participated in the discussion. Legel described the exam development process in detail. The City, he recounted, had set the “parameters” for the exams, specifically, the requirement of written and oral components with a 60/40 weighting. Id., at A923, A974. For security reasons, Department officials had not been permitted to check the content of the questions prior to their administration. Instead, IOS retained a senior fire officer from Georgia to review the exams “for con*615tent and fidelity to the source material.” Id., at A936. Legel defended the exams as “facially neutral,” and stated that he “would stand by the[ir] validity.” Id., at A962. City officials did not dispute the neutrality of IOS’s work. But, they cautioned, even if individual exam questions had no intrinsic bias, the selection process as a whole may nevertheless have been deficient. The officials urged the CSB to consult with experts about the “larger picture.” Id., at A1012.
At its fourth meeting, CSB solicited the views of three individuals with testing-related expertise. Dr. Christopher Hornick, an industrial/organizational psychology consultant with 25 years’ experience with police and firefighter testing, described the exam results as having “relatively high adverse impact.” Id., at A1028. Most of the tests he had developed, Hornick stated, exhibited “significantly and dramatically less adverse impact.” Id., at A1029. Hornick downplayed the notion of “facial neutrality.” It was more important, he advised the CSB, to consider “the broader issue of how your procedures and your rules and the types of tests that you are using are contributing to the adverse impact.” Id., at A1038.
Specifically, Hornick questioned New Haven’s union-prompted 60/40 written/oral examination structure, noting the availability of “different types of testing procedures that are much more valid in terms of identifying the best potential supervisors in [the] fire department.” Id., at A1032. He suggested, for example, “an assessment center process, which is essentially an opportunity for candidates ... to demonstrate how they would address a particular problem as opposed to just verbally saying it or identifying the correct option on a written test.” Id., at A1039-A1040. Such selection processes, Hornick said, better “identify] the best possible people” and “demonstrate dramatically less adverse impacts.” Ibid. Hornick added:
*616“I’ve spoken to at least 10,000, maybe 15,000, firefighters in group settings in my consulting practice and I have never one time ever had anyone in the fire service say to me, ‘Well, the person who answers — gets the highest score on a written job knowledge, multiple-guess test makes the best company officer.’ We know that it’s not as valid as other procedures that exist.” Id., at A1033.
See also id., at A1042-A1043 (“I think a person’s leadership skills, their command presence, their interpersonal skills, their management skills, their tactical skills could have been identified and evaluated in a much more appropriate way.”).
Hornick described the written test itself as “reasonably good,” id., at A1041, but he criticized the decision not to allow Department officials to check the content. According to Hornick, this “inevitably” led to “test[ing] for processes and procedures that don’t necessarily match up into the department.” Id., at A1034-A1035. He preferred “experts from within the department who have signed confidentiality agreements ... to make sure that the terminology and equipment that’s being identified from standardized reading sources apply to the department.” Id., at A1035.
Asked whether he thought the City should certify the results, Hornick hedged: “There is adverse impact in the test. That will be identified in any proceeding that you have. You will have industrial psychology experts, if it goes to court, on both sides. And it will not be a pretty or comfortable position for anyone to be in.” Id., at A1040-A1041. Perhaps, he suggested, New Haven might certify the results but immediately begin exploring “alternative ways to deal with these issues” in the future. Id., at A1041.
The two other witnesses made relatively brief appearances. Vincent Lewis, a specialist with the Department of Homeland Security and former fire officer in Michigan, believed the exams had generally tested relevant material, although he noted a relatively heavy emphasis on questions *617pertaining to being an “apparatus driver.” He suggested that this may have disadvantaged test takers “who had not had the training or had not had an opportunity to drive the apparatus.” Id., at A1051. He also urged the CSB to consider whether candidates had, in fact, enjoyed equal access to the study materials. Ibid. Cf. supra, at 613-614.
Janet Helms, a professor of counseling psychology at Boston College, observed that two-thirds of the incumbent fire officers who submitted job analyses to IOS during the exam-design phase were Caucasian. Members of different racial groups, Helms told the CSB, sometimes do their jobs in different ways, “often because the experiences that are open to white male firefighters are not open to members of these other under-represented groups.” CA2 App. A1063-A1064. The heavy reliance on job analyses from white firefighters, she suggested, may thus have introduced an element of bias. Id., at A1063.
The CSB’s fifth and final meeting began with statements from City officials recommending against certification. Ude, New Haven’s counsel, repeated the applicable disparate-impact standard:
“[A] finding of adverse impact is the beginning, not the end, of a review of testing procedures. Where a procedure demonstrates adverse impact, you look to how closely it is related to the job that you’re looking to fill and you also look at whether there are other ways to test for those qualities, those traits, those positions that are equally valid with less adverse impact.” Id., at A1100-A1101.
New Haven, Ude and other officials asserted, would be vulnerable to Title VII liability under this standard. Even if the exams were “facially neutral,” significant doubts had been raised about whether they properly assessed the key attributes of a successful fire officer. Id., at A1103. See also id., at A1125 (“Upon close reading of the exams, the *618questions themselves would appear to test a candidate’s ability to memorize textbooks but not necessarily to identify solutions to real problems on the fire ground.”). Moreover, City officials reminded the CSB, Hornick and others had identified better, less discriminatory selection methods— such as assessment centers or exams with a more heavily weighted oral component. Id., at A1108-A1109, A1129-A1130.
After giving members of the public a final chance to weigh in, the CSB voted on certification, dividing 2 to 2. By rule, the result was noncertification. Voting no, Commissioner Webber stated, “I originally was going to vote to certify. . . . But I’ve heard enough testimony here to give me great doubts about the test itself and . . . some of the procedures. And I believe we can do better.” Id., at A1157. Commissioner Tirado likewise concluded that the “flawed” testing process counseled against certification. Id., at A1158. Chairman Segaloff and Commissioner Caplan voted to certify. According to Segaloff, the testimony had not “compelled [him] to say this exam was not job-related,” and he was unconvinced that alternative selection processes would be “less discriminatory.” Id., at A1159-A1160. Both Segaloff and Caplan, however, urged the City to undertake civil-service reform. Id., at A1150-A1154.
C
Following the CSB’s vote, petitioners — 17 white firefighters and one Hispanic firefighter, all of whom had high marks on the exams — filed suit in the United States District Court for the District of Connecticut. They named as defendants — respondents here — the City, several City officials, a local political activist, and the two CSB members who voted against certifying the results. By opposing certification, petitioners alleged, respondents had discriminated against them in violation of Title VII’s disparate-treatment provision and the Fourteenth Amendment’s Equal Protec*619tion Clause. The decision not to certify, respondents answered, was a lawful effort to comply with Title VII’s disparate-impact provision and thus could not have run afoul of Title VII’s prohibition of disparate treatment. Characterizing respondents’ stated rationale as a mere pretext, petitioners insisted that New Haven would have had a solid defense to any disparate-impact suit.
In a decision summarily affirmed by the Court of Appeals, the District Court granted summary judgment for respondents. 554 F. Supp. 2d 142 (Conn. 2006), aff’d, 530 F. 3d 87 (CA2 2008) (per curiam). Under Second Circuit precedent, the District Court explained, “the intent to remedy the disparate impact” of a promotional exam “is not equivalent to an intent to discriminate against non-minority applicants.” 554 F. Supp. 2d, at 157 (quoting Hayden v. County of Nassau, 180 F. 3d 42, 51 (CA2 1999)). Rejecting petitioners’ pretext argument, the court observed that the exam results were sufficiently skewed “to make out a prima facie case of discrimination” under Title VII’s disparate-impact provision. 554 F. Supp. 2d, at 158. Had New Haven gone forward with certification and been sued by aggrieved minority test takers, the City would have been forced to defend tests that were presumptively invalid. And, as the CSB testimony of Hornick and others indicated, overcoming that presumption would have been no easy task. Id., at 153-156. Given Title VII’s preference for voluntary compliance, the court held, New Haven could lawfully discard the disputed exams even if the City had not definitively “pinpoint[ed]” the source of the disparity and “ha[d] not yet formulated a better selection method.” Id., at 156.
Respondents were no doubt conscious of race during their decisionmaking process, the court acknowledged, but this did not mean they had engaged in racially disparate treatment. The conclusion they had reached and the action thereupon taken were race neutral in this sense: “[A]ll the test results were discarded, no one was promoted, and firefighters of *620every race will have to participate in another selection process to be considered for promotion.” Id., at 158. New Haven’s action, which gave no individual a preference, “was ‘simply not analogous to a quota system or a minority set-aside where candidates, on the basis of their race, are not treated uniformly.’ ” Id., at 157 (quoting Hayden, 180 F. 3d, at 50). For these and other reasons, the court also rejected petitioners’ equal protection claim.
II
A
Title VII became effective in July 1965. Employers responded to the law by eliminating rules and practices that explicitly barred racial minorities from “white” jobs. But removing overtly race-based job classifications did not usher in genuinely equal opportunity. More subtle — and sometimes unconscious — forms of discrimination replaced once undisguised restrictions.
In Griggs v. Duke Power Co., 401 U. S. 424 (1971), this Court responded to that reality and supplied important guidance on Title VII’s mission and scope. Congress, the landmark decision recognized, aimed beyond “disparate treatment”; it targeted “disparate impact” as well. Title VII’s original text, it was plain to the Court, “proscribe [d] not only overt discrimination but also practices that are fair in form, but discriminatory in operation.” Id., at 431.2 Only by ig*621noring Griggs could one maintain that intentionally disparate treatment alone was Title VIPs “original, foundational prohibition,” and disparate impact a mere afterthought. Cf. ante, at 581.
Griggs addressed Duke Power Company’s policy that applicants for positions, save in the company’s labor department, be high school graduates and score satisfactorily on two professionally prepared aptitude tests. “[T]here was no showing of a discriminatory purpose in the adoption of the diploma and test requirements.” 401 U. S., at 428. The policy, however, “operated to render ineligible a markedly disproportionate number of [African-Americans].” Id., at 429. At the time of the litigation, in North Carolina, where the Duke Power plant was located, 34 percent of white males, but only 12 percent of African-American males, had high school diplomas. Id., at 430, n. 6. African-Americans also failed the aptitude tests at a significantly higher rate than whites. Ibid. Neither requirement had been “shown to bear a demonstrable relationship to successful performance of the jobs for which it was used.” Id., at 431.
The Court unanimously held that the company’s diploma and test requirements violated Title VII. “[T]o achieve equality of employment opportunities,” the Court comprehended, Congress “directed the thrust of the Act to the consequences of employment practices, not simply the motivation.” Id., at 429, 432. That meant “unnecessary barriers to employment” must fall, even if “neutral on their face” and “neutral in terms of intent.” Id., at 430, 431. “The touchstone” for determining whether a test or qualification meets Title VIPs measure, the Court said, is not “good intent or the absence of discriminatory intent”; it is “business necessity.” Id., at 431, 432. Matching procedure to substance, the Griggs Court observed, Congress “placed on the em*622ployer the burden of showing that any given requirement... ha[s] a manifest relationship to the employment in question.” Id., at 432.
In Albemarle Paper Co. v. Moody, 422 U. S. 405 (1975), the Court, again without dissent, elaborated on Griggs. When an employment test “select[s] applicants for hire or promotion in a racial pattern significantly different from the pool of applicants,” the Court reiterated, the employer must demonstrate a “manifest relationship” between test and job. 422 U. S., at 425. Such a showing, the Court cautioned, does not necessarily mean the employer prevails: “[I]t remains open to the complaining party to show that other tests or selection devices, without a similarly undesirable racial effect, would also serve the employer’s legitimate interest in ‘efficient and trustworthy workmanship.’” Ibid.
Federal trial and appellate courts applied Griggs and Albemarle to disallow a host of hiring and promotion practices that “operate[d] as ‘built in headwinds’ for minority groups.” Griggs, 401 U. S., at 432. Practices discriminatory in effect, courts repeatedly emphasized, could be maintained only upon an employer’s showing of “an overriding and compelling business purpose.” Chrisner v. Complete Auto Transit, Inc., 645 F. 2d 1251, 1261, n. 9 (CA6 1981).3 That a prac*623tice served “legitimate management functions” did not, it was generally understood, suffice to establish business necessity. Williams v. Colorado Springs, Colo., School Dish, 641 F. 2d 835, 840-841 (CA10 1981) (internal quotation marks omitted). Among selection methods cast aside for lack of a “manifest relationship” to job performance were a number of written hiring and promotional examinations for firefighters.4
Moving in a different direction, in Wards Cove Packing Co. v. Atonio, 490 U. S. 642 (1989), a bare majority of this Court significantly modified the Griggs-Albemarle delineation of Title VIPs disparate-impact proscription. As to business necessity for a practice that disproportionately excludes members of minority groups, Wards Cove held, the employer bears only the burden of production, not the burden of persuasion. 490 U. S., at 659-660. And in place of the instruction that the challenged practice “must have a manifest relationship to the employment in question,” Griggs, 401 U. S., at 432, Wards Cove said that the practice would be permissible as long as it “serve[d], in a significant way, the legitimate employment goals of the employer,” 490 U. S., at 659.
*624In response to Wards Cove and “a number of [other] recent decisions by the United States Supreme Court that sharply cut back on the scope and effectiveness of [civil rights] laws,” Congress enacted the Civil Rights Act of 1991. H. R. Rep. No. 102-40, pt. 2, p. 2 (1991). Among the 1991 alterations, Congress formally codified the disparate-impact component of Title VII. In so amending the statute, Congress made plain its intention to restore “the concepts of ‘business necessity’ and ‘job related’ enunciated by the Supreme Court in Griggs v. Duke Power Co. . . . and in other Supreme Court decisions prior to Wards Cove Packing Co. v. Atonio.” §3(2), 105 Stat. 1071. Once a complaining party demonstrates that an employment practice causes a disparate impact, amended Title VII states, the burden is on the employer “to demonstrate that the challenged practice is job related for the position in question and consistent with business necessity.” 42 U. S. C. §2000e-2(k)(l)(A)(i). If the employer carries that substantial burden, the complainant may respond by identifying “an alternative employment practice” which the employer “refuses to adopt.” § 2000e-2(k)(l)(A)(ii), (C).
B
Neither Congress’ enactments nor this Court’s Title VII precedents (including the now-discredited decision in Wards Cove) offer even a hint of “conflict” between an employer’s obligations under the statute’s disparate-treatment and disparate-impact provisions. Cf. ante, at 580. Standing on an equal footing, these twin pillars of Title VII advance the same objectives: ending workplace discrimination and promoting genuinely equal opportunity. See McDonnell Douglas Cory. v. Green, 411 U. S. 792, 800 (1973).
Yet the Court today sets at odds the statute’s core directives. When an employer changes an employment practice in an effort to comply with Title VII’s disparate-impact pro*625vision, the Court reasons, it acts “because of race” — something Title VIPs disparate-treatment provision, see § 2000e-2(a)(1), generally forbids. Ante, at 579-580. This characterization of an employer’s compliance-directed action shows little attention to Congress’ design or to the Griggs line of cases Congress recognized as pathmarking.
“[O]ur task in interpreting separate provisions of a single Act is to give the Act the most harmonious, comprehensive meaning possible in light of the legislative policy and purpose.” Weinberger v. Hynson, Westcott & Dunning, Inc., 412 U. S. 609, 631-632 (1973) (internal quotation marks omitted). A particular phrase need not “extend to the outer limits of its definitional possibilities” if an incongruity would result. Dolan v. Postal Service, 546 U. S. 481, 486 (2006). Here, Title VII’s disparate-treatment and disparate-impact proscriptions must be read as complementary.
In codifying the Griggs and Albemarle instructions, Congress declared unambiguously that selection criteria operating to the disadvantage of minority group members can be retained only if justified by business necessity.5 In keeping with Congress’ design, employers who reject such criteria due to reasonable doubts about their reliability can hardly be held to have engaged in discrimination “because of” race. A reasonable endeavor to comply with the law and to ensure that qualified candidates of all races have a fair opportunity to compete is simply not what Congress meant to interdict. I would therefore hold that an employer who jettisons a selection device when its disproportionate racial impact becomes apparent does not violate Title VII’s disparate-treatment bar automatically or at all, subject to this key condition: The employer must have good cause to believe the *626device would not withstand examination for business necessity. Cf. Faragher v. Boca Raton, 524 U. S. 775, 806 (1998) (observing that it accords with “clear statutory policy” for employers “to prevent violations” and “make reasonable efforts to discharge their duty” under Title VII).
EEOC’s interpretative guidelines are corroborative. “[B]y the enactment of title VII,” the guidelines state, “Congress did not intend to expose those who comply with the Act to charges that they are violating the very statute they are seeking to implement.” 29 CFR § 1608.1(a) (2008). Recognizing EEOC’s “enforcement responsibility” under Title VII, we have previously accorded the Commission’s position respectful consideration. See, e.g., Albemarle, 422 U. S., at 431; Griggs, 401 U. S., at 434. Yet the Court today does not so much as mention EEOC’s counsel.
Our precedents defining the contours of Title VII’s disparate-treatment prohibition further confirm the absence of any intrastatutory discord. In Johnson v. Transportation Agency, Santa Clara Cty., 480 U. S. 616 (1987), we upheld a municipal employer’s voluntary affirmative-action plan against a disparate-treatment challenge. Pursuant to the plan, the employer selected a woman for a road-dispatcher position, a job category traditionally regarded as “male.” A male applicant who had a slightly higher interview score brought suit under Title VII. This Court rejected his claim and approved the plan, which allowed consideration of gender as “one of numerous factors.” Id., at 638. Such consideration, we said, is “fully consistent with Title VII” because plans of that order can aid “in eliminating the vestiges of discrimination in the workplace.” Id., at 642.
This litigation does not involve affirmative action. But if the voluntary affirmative action at issue in Johnson does not discriminate within the meaning of Title VII, neither does an employer’s reasonable effort to comply with Title VII’s disparate-impact provision by refraining from action of doubtful consistency with business necessity.
*627c
To “reconcile” the supposed “conflict” between disparate treatment and disparate impact, the Court offers an enigmatic standard. Ante, at 580. Employers may attempt to comply with Title VII’s disparate-impact provision, the Court declares, only where there is a “strong basis in evidence” documenting the necessity of their action. Ante, at 583. The Court’s standard, drawn from inapposite equal protection precedents, is not elaborated. One is left to wonder what cases would meet the standard and why the Court is so sure cases of this genre do not.
1
In construing Title VII, I note preliminarily, equal protection doctrine is of limited utility. The Equal Protection Clause, this Court has held, prohibits only intentional discrimination; it does not have a disparate-impact component. See Personnel Administrator of Mass. v. Feeney, 442 U. S. 256, 272 (1979); Washington v. Davis, 426 U. S. 229, 239 (1976). Title VII, in contrast, aims to eliminate all forms of employment discrimination, unintentional as well as deliberate. Until today, cf. ante, at 584; ante, p. 594 (Scalia, J., concurring), this Court has never questioned the constitutionality of the disparate-impact component of Title VII, and for good reason. By instructing employers to avoid needlessly exclusionary selection processes, Title VIPs disparate-impact provision calls for a “race-neutral means to increase minority . . . participation” — something this Court’s equal protection precedents also encourage. See Adarand Constructors, Inc. v. Pena, 515 U. S. 200, 238 (1995) (quoting Richmond v. J. A. Croson Co., 488 U. S. 469, 507 (1989)). “The very radicalism of holding disparate impact doctrine unconstitutional as a matter of equal protection,” moreover, “suggests that only a very uncompromising court would issue such a decision.” Primus, Equal Protection and Dis*628parate Impact: Round Three, 117 Harv. L. Rev. 493, 585 (2003).
The cases from which the Court draws its strong-basis-in-evidence standard are particularly inapt; they concern the constitutionality of absolute racial preferences. See Wygant v. Jackson Bd. of Ed., 476 U. S. 267, 277 (1986) (plurality opinion) (invalidating a school district’s plan to lay off nonminority teachers while retaining minority teachers with less seniority); Croson, 488 U. S., at 499-500 (rejecting a set-aside program for minority contractors that operated as “an unyielding racial quota”). An employer’s effort to avoid Title VII liability by repudiating a suspect selection method scarcely resembles those cases. Race was not merely a relevant consideration in Wygant and Croson; it was the decisive factor. Observance of Title VII’s disparate-impact provision, in contrast, calls for no racial preference, absolute or otherwise. The very purpose of the provision is to ensure that individuals are hired and promoted based on qualifications manifestly necessary to successful performance of the job in question, qualifications that do not screen out members of any race.6
2
The Court’s decision in this litigation underplays a dominant Title VII theme. This Court has repeatedly emphasized that the statute “should not be read to thwart” efforts at voluntary compliance. Johnson, 480 U. S., at 630. Such *629compliance, we have explained, is “the preferred means of achieving [Title VIPs] objectives.” Firefighters v. Cleveland, 478 U. S. 501, 515 (1986). See also Kolstad v. American Dental Assn., 527 U. S. 526, 545 (1999) (“Dissuading employers from [taking voluntary action] to prevent discrimination in the workplace is directly contrary to the purposes underlying Title VII.”); 29 CFR § 1608.1(e). The strong-basis-in-evidence standard, however, as barely described in general, and cavalierly applied in this litigation, makes voluntary compliance a hazardous venture.
As a result of today’s decision, an employer who discards a dubious selection process can anticipate costly disparate-treatment litigation in which its chances for success — even for surviving a summary-judgment motion — are highly problematic. Concern about exposure to disparate-impact liability, however well grounded, is insufficient to insulate an employer from attack. Instead, the employer must make a “strong” showing that (1) its selection method was “not job related and consistent with business necessity,” or (2) that it refused to adopt “an equally valid, less discriminatory alternative.” Ante, at 587. It is hard to see how these requirements differ from demanding that an employer establish “a provable, actual violation” against itself. Cf. ante, at 583. There is indeed a sharp conflict here, but it is not the false one the Court describes between Title VIPs core provisions. It is, instead, the discordance of the Court’s opinion with the voluntary compliance ideal. Cf. Wygant, 476 U. S., at 290 (O’Connor, J., concurring in part and concurring in judgment) (“The imposition of a requirement that public employers make findings that they have engaged in illegal discrimination before they [act] would severely undermine public employers’ incentive to meet voluntarily their civil rights obligations.”).7
*6303
The Court’s additional justifications for announcing a strong-basis-in-evidence standard are unimpressive. First, discarding the results of tests, the Court suggests, calls for a heightened standard because it “upset[s] an employee’s legitimate expectation.” Ante, at 585. This rationale puts the cart before the horse. The legitimacy of an employee’s expectation depends on the legitimacy of the selection method. If an employer reasonably concludes that an exam fails to identify the most qualified individuals and needlessly shuts out a segment of the applicant pool, Title VII surely does not compel the employer to hire or promote based on the test, however unreliable it may be. Indeed, the statute’s prime objective is to prevent exclusionary practices from “operating] to ‘freeze’ the status quo.” Griggs, 401 U. S., at 430.
Second, the Court suggests, anything less than a strong-basis-in-evidence standard risks creating “a de facto quota system, in which ... an employer could discard test results . . . with the intent of obtaining the employer’s preferred racial balance.” Ante, at 581-582. Under a reasonableness standard, however, an employer could not cast aside a selection method based on a statistical disparity alone.8 The employer must have good cause to believe that the method *631screens out qualified applicants and would be difficult to justify as grounded in business necessity. Should an employer repeatedly reject test results, it would be fair, I agree, to infer that the employer is simply seeking a racially balanced outcome and is not genuinely endeavoring to comply with Title VII.
The Court stacks the deck further by denying respondents any chance to satisfy the newly announced strong-basis-in-evidence standard. When this Court formulates a new legal rule, the ordinary course is to remand and allow the lower courts to apply the rule in the first instance. See, e. g., Johnson v. California, 543 U. S. 499, 515 (2005); Pullman-Standard v. Swint, 456 U. S. 273, 291 (1982). I see no good reason why the Court fails to follow that course today. Indeed, the sole basis for the Court’s peremptory ruling is the demonstrably false pretension that respondents showed “nothing more” than “a significant statistical disparity.” Ante, at 587; see supra, at 630, n. 8.9
*632III
A
Applying what I view as the proper standard to the record thus far made, I would hold that New Haven had ample cause to believe its selection process was flawed and not justified by business necessity. Judged by that standard, petitioners have not shown that New Haven’s failure to certify the exam results violated Title VII’s disparate-treatment provision.10
The City, all agree, “was faced with a prima facie ease of disparate-impact liability,” ante, at 586 (majority opinion): The pass rate for minority candidates was half the rate for nonminority candidates, and virtually no minority candidates would have been eligible for promotion had the exam results been certified. Alerted to this stark disparity, the CSB heard expert and lay testimony, presented at public hearings, in an endeavor to ascertain whether the exams were fair and consistent with business necessity. Its investigation revealed grave cause for concern about the exam process itself and the City’s failure to consider alternative selection devices.
Chief among the City’s problems was the very nature of the tests for promotion. In choosing to use written and oral exams with a 60/40 weighting, the City simply adhered to the union’s preference and apparently gave no consideration to whether the weighting was likely to identify the most qualified fire-officer candidates.11 There is strong reason to think it was not.
*633Relying heavily on written tests to select fire officers is a questionable practice, to say the least. Successful fire officers, the City’s description of the position makes clear, must have the “[a]bility to lead personnel effectively, maintain discipline, promote harmony, exercise sound judgment, and cooperate with other officials.” CA2 App. A432. These qualities are not well measured by written tests. Testifying before the CSB, Christopher Hornick, an exam-design expert with more than two decades of relevant experience, was emphatic on this point: Leadership skills, command presence, and the like “could have been identified and evaluated in a much more appropriate way.” Id., at A1042-A1043.
Hornick’s commonsense observation is mirrored in case law and in Title VII’s administrative guidelines. Courts have long criticized written firefighter promotion exams for being “more probative of the test taker’s ability to recall what a particular text stated on a given topic than of his firefighting or supervisory knowledge and abilities.” Vulcan Pioneers, Inc. v. New Jersey Dept. of Civil Serv., 625 F. Supp. 527, 539 (NJ 1985). A fire officer’s job, courts have *634observed, “involves complex behaviors, good interpersonal skills, the ability to make decisions under tremendous pressure, and a host of other abilities — none of which is easily measured by a written, multiple choice test.” Firefighters Inst. for Racial Equality v. St. Louis, 616 F. 2d 350, 359 (CA8 1980).12 Interpreting the Uniform Guidelines, EEOC and other federal agencies responsible for enforcing equal opportunity employment laws have similarly recognized that, as measures of “interpersonal relations” or “ability to function under danger (e. g., firefighters),” “[p]encil-andpaper tests . . . generally are not close enough approximations of work behaviors to show content validity.” 44 Fed. Reg. 12007 (1979). See also 29 CFR § 1607.15(C)(4).13
Given these unfavorable appraisals, it is unsurprising that most municipal employers do not evaluate their fire-officer candidates as New Haven does. Although comprehensive statistics are scarce, a 1996 study found that nearly two-thirds of surveyed municipalities used assessment centers *635(“simulations of the real world of work”) as part of their promotion processes. P. Lowry, A Survey of the Assessment Center Process in the Public Sector, 25 Public Personnel Management 307, 315 (1996). That figure represented a marked increase over the previous decade, see ibid., so the percentage today may well be even higher. Among municipalities still relying in part on written exams, the median weight assigned to them was 30 percent — half the weight given to New Haven’s written exam. Id., at 309.
Testimony before the CSB indicated that these alternative methods were both more reliable and notably less discriminatory in operation. According to Donald Day of the International Association of Black Professional Firefighters, nearby Bridgeport saw less skewed results after switching to a selection process that placed primary weight on an oral exam. CA2 App. A830-A832; see supra, at 614. And Hornick described assessment centers as “demonstrat[ing] dramatically less adverse impacts” than written exams. CA2 App. A1040.14 Considering the prevalence of these proven alternatives, New Haven was poorly positioned to argue that promotions based on its outmoded and exclusionary selection process qualified as a business necessity. Cf. Robinson v. Lorillard Corp., 444 F. 2d 791, 798, n. 7 (CA4 1971) (“It should go without saying that a practice is hardly ‘necessary’ if an alternative practice better effectuates its intended purpose or is equally effective but less discriminatory.”).15
*636Ignoring the conceptual and other defects in New Haven’s selection process, the Court describes the exams as “painstaking[ly]” developed to test “relevant” material and on that basis finds no substantial risk of disparate-impact liability. See ante, at 588. Perhaps such reasoning would have sufficed under Wards Cove, which permitted exclusionary practices as long as they advanced an employer’s “legitimate” goals. 490 U. S., at 659. But Congress repudiated Wards Cove and reinstated the “business necessity” rule attended by a “manifest relationship” requirement. See Griggs, 401 U. S., at 431-432. See also supra, at 624. Like the chess player who tries to win by sweeping the opponent’s pieces off the table, the Court simply shuts from its sight the formidable obstacles New Haven would have faced in defending against a disparate-impact suit. See Lanning v. Southeastern Pa. Transp. Auth., 181 F. 3d 478, 489 (CA3 1999) (“Judicial application of a standard focusing solely on whether the qualities measured by an . . . exam bear some relationship to the job in question would impermissibly write out the business necessity prong of the Act’s chosen standard.”).
*637That IOS representative Chad Legel and his team may-have been diligent in designing the exams says little about the exams’ suitability for selecting fire officers. IOS worked within the City’s constraints. Legel never discussed with the City the propriety of the 60/40 weighting and “was not asked to consider the possibility of an assessment center.” CA2 App. A522. See also id., at A467.' The IOS exams, Legel admitted, had not even attempted to assess “command presence”: “[Y]ou would probably be better off with an assessment center if you cared to measure that.” Id., at A521. Cf. Boston Chapter, NAACP, Inc. v. Beecher, 504 F. 2d 1017, 1021-1022 (CA1 1974) (“A test fashioned from materials pertaining to the job . . . superficially may seem job-related. But what is at issue is whether it demonstrably selects people who will perform better the required on-the-job behaviors.”).
In addition to the highly questionable character of the exams and the neglect of available alternatives, the City had other reasons to worry about its vulnerability to disparate-impact liability. Under the City’s ground rules, IOS was not allowed to show the exams to anyone in the New Haven Fire Department prior to their administration. This “precluded [IOS] from being able to engage in [its] normal subject matter expert review process” — something Legel described as “very critical.” CA2 App. A477, A506. As a result, some of the exam questions were confusing or irrelevant, and the exams may have overtested some subject-matter areas while missing others. See, e. g., id., at A1034-A1035, A1051. Testimony before the CSB also raised questions concerning unequal access to study materials, see id., at A857-A861, and the potential bias introduced by relying principally on job analyses from nonminority fire officers to develop the exams, see id., at A1063-A1064.16 See also supra, at 613-614, 617.
*638The Court criticizes New Haven for failing to obtain a “technical report” from IOS, which, the Court maintains, would have provided “detailed information to establish the validity of the exams.” Ante, at 589. The record does not substantiate this assertion. As Legel testified during his deposition, the technical report merely summarized “the steps that [IOS] took methodologically speaking,” and would not have established the exams’ reliability. CA2 App. A461. See also id., at A462 (the report “doesn’t say anything that other documents that already existed wouldn’t say”).
In sum, the record solidly establishes that the City had good cause to fear disparate-impact liability. Moreover, the Court supplies no tenable explanation why the evidence of the tests’ multiple deficiencies does not create at least a triable issue under a strong-basis-in-evidence standard.
B
Concurring in the Court’s opinion, Justice Alito asserts that summary judgment for respondents would be improper even if the City had good cause for its noncertification decision. A reasonable jury, he maintains, could have found that respondents were not actually motivated by concern about disparate-impact litigation, but instead sought only “to placate a politically important [African-American] constitu*639ency.” Ante, at 597. As earlier noted, I would not oppose a remand for further proceedings fair to both sides. See supra, at 632, n. 10. It is the Court that has chosen to short circuit this litigation based on its pretension that the City has shown, and can show, nothing more than a statistical disparity. See supra, at 630, n. 8, 631. Justice Alito compounds the Court’s error.
Offering a truncated synopsis of the many hours of deliberations undertaken by the CSB, Justice Alito finds evidence suggesting that respondents’ stated desire to comply with Title YII was insincere, a mere “pretext” for discrimination against white firefighters. Ante, at 596-597. In support of his assertion, Justice Alito recounts at length the alleged machinations of Rev. Boise Kimber (a local political activist), Mayor John DeStefano, and certain members of the mayor’s staff. See ante, at 598-604.
Most of the allegations Justice Alito repeats are drawn from petitioners’ statement of facts they deem undisputed, a statement displaying an adversarial zeal not uncommonly found in such presentations.17 What cannot credibly be de*640nied, however, is that the decision against certification of the exams was made neither by Kimber nor by the mayor and his staff. The relevant decision was made by the CSB, an unelected, politically insulated body. It is striking that Justice Alito’s concurrence says hardly a word about the CSB itself, perhaps because there is scant evidence that its motivation was anything other than to comply with Title VIPs disparate-impact provision. Notably, petitioners did not even seek to take depositions of the two commissioners who voted against certification. Both submitted uncontested affidavits declaring unequivocally that their votes were “based solely on [their] good faith belief that certification” would have discriminated against minority candidates in violation of federal law. CA2 App. A1605, A1611.
Justice Alito discounts these sworn statements, suggesting that the CSB’s deliberations were tainted by the preferences of Kimber and City officials, whether or not the CSB itself was aware of the taint. Kimber and City officials, Justice Alito speculates, decided early on to oppose certification and then “engineered” a skewed presentation to the CSB to achieve their preferred outcome. Ante, at 606.
As an initial matter, Justice Alito exaggerates the influence of these actors. The CSB, the record reveals, designed and conducted an inclusive decisionmaking process, in which it heard from numerous individuals on both sides of the certification question. See, e.g., CA2 App. A1090. Kimber and others no doubt used strong words to urge the CSB not to certify the exam results, but the CSB received “pressure” from supporters of certification as well as opponents. Cf. ante, at 600. Petitioners, for example, engaged counsel to speak on their behalf before the CSB. Their counsel did not mince words: “[I]f you discard these results,” she warned, “you will get sued. You will force the taxpay*641ers of the city of New Haven into protracted litigation.” CA2 App. A816. See also id., at A788.
The local firefighters union — an organization required by law to represent all the City’s firefighters — was similarly outspoken in favor of certification. Discarding the test results, the union’s president told the CSB, would be “totally ridiculous.” Id., at A806. He insisted, inaccurately, that the City was not at risk of disparate-impact liability because the exams were administered pursuant to “a collective bargaining agreement.” Id., at A1137. Cf. supra, at 632-633, n. 11. Never mentioned by Justice Alito in his attempt to show testing expert Christopher Hornick’s alliance with the City, ante, at 603-604, the CSB solicited Hornick’s testimony at the union’s suggestion, not the City’s. CA2 App. A1128. Hornick’s cogent testimony raised substantial doubts about the exams’ reliability. See supra, at 615-616.18
There is scant cause to suspect that maneuvering or overheated rhetoric, from either side, prevented the CSB from evenhandedly assessing the reliability of the exams and rendering an independent, good-faith decision on certification. Justice Alito acknowledges that the CSB had little patience for Kimber’s antics. Ante, at 600-602.19 As to petitioners, Chairman Segaloff — who voted to certify the exam *642results — dismissed the threats made by their counsel as unhelpful and needlessly “inflammatory.” CA2 App. A821. Regarding the views expressed by City officials, the CSB made clear that they were entitled to no special weight. Id., at A1080.20
In any event, Justice Alito’s analysis contains a more fundamental flaw: It equates political considerations with unlawful discrimination. As Justice Alito sees it, if the mayor and his staff were motivated by their desire “to placate a . . . racial constituency,” ante, at 597, then they engaged in unlawful discrimination against petitioners. But Justice Alito fails to ask a vital question: “[P]lacate” how? That political officials would have politics in mind is hardly extraordinary, and there are many ways in which a politician can attempt to win over a constituency — including a racial constituency — without engaging in unlawful discrimination. As courts have recognized, “[politicians routinely respond to bad press . . . , but it is not a violation of Title VII to take advantage of a situation to gain political favor.” Henry v. Jones, 507 F. 3d 558, 567 (CA7 2007).
The real issue, then, is not whether the mayor and his staff were politically motivated; it is whether their attempt to score political points was legitimate (1 e., nondiscriminatory). Were they seeking to exclude white firefighters from promotion (unlikely, as a fair test would undoubtedly result in the addition of white firefighters to the officer ranks), or did they realize, at least belatedly, that their tests could be toppled in a disparate-impact suit? In the latter case, *643there is no disparate-treatment violation. Justice Alito, I recognize, would disagree. In his view, an employer^ action to avoid Title VII disparate-impact liability qualifies as a presumptively improper race-based employment decision. See ante, at 597. I reject that construction of Title VII. See supra, at 625-627. As I see it, when employers endeavor to avoid exposure to disparate-impact liability, they do not thereby encounter liability for disparate treatment.
Applying this understanding of Title VII, supported by Griggs and the long line of decisions following Griggs, see supra, at 623-624, and nn. 3-4, the District Court found no genuine dispute of material fact. That court noted, particularly, the guidance furnished by Second Circuit precedent. See supra, at 619. Petitioners’ allegations that City officials took account of politics, the District Court determined, simply “d[id] not suffice” to create an inference of unlawful discrimination. 554 F. Supp. 2d, at 160, n. 12. The noncertification decision, even if undertaken “in a political context,” reflected a legitimate “intent not to implement a promotional process based on testing results that had an adverse impact.” Id., at 158, 160. Indeed, the District Court perceived “a total absence of any evidence of discriminatory animus towards [petitioners].” Id., at 158. See also id., at 162 (“Nothing in the record in this case suggests that the City defendants or CSB acted ‘because of’ discriminatory animus toward [petitioners] or other non-minority applicants for promotion.”). Perhaps the District Court could have been more expansive in its discussion of these issues, but its conclusions appear entirely consistent with the record before it.21
*644It is indeed regrettable that the City’s noncertification decision would have required all candidates to go through another selection process. But it would have been more regrettable to rely on'flawed exams to shut out candidates who may well have the command presence and other qualities needed to excel as fire officers. Yet that is the choice the Court makes today. It is a choice that breaks the promise of Griggs that groups long denied equal opportunity would not be held back by tests “fair in form, but discriminatory in operation.” 401 U. S., at 431.
* * *
These cases present an unfortunate situation, one New Haven might well have avoided had it utilized a better selection process in the first place. But what this litigation does not present is race-based discrimination in violation of Title VIL I dissent from the Court’s judgment, which rests on the false premise that respondents showed “a significant statistical disparity,” but “nothing more.” See ante, at 587.
Although the dissent disputes it, see post, at 639-640, n. 17, the record certainly permits the inference that petitioners’ allegation is true. See App. to Pet. for Cert. in No. 07-1428, pp. 846a-851a (deposition of Dubois-Walton).
The City’s heavy reliance on Hornick’s testimony makes the two chiefs’ silence all the more striking. See supra, at 599-600. While Hornick knew little or nothing about the tests he criticized, the two chiefs were involved “during the lengthy process that led to the devising of the administration of these exams,” App. to Pet. for Cert. in No. 07-1428, at 847a, including “collaborating with City officials on the extensive job analyses that were done,” “selection of the oral panelists,” and selection of “the proper content and subject matter of the exams,” id., at 847a-848a.
Never mind the flawed tests New Haven used and the better selection methods used elsewhere, Justice Alito’s concurring opinion urges. Overriding all else, racial politics, fired up by a strident African-American pastor, were at work in New Haven. See ante, at 599-604. Even a detached and disinterested observer, however, would have every reason to ask: Why did such racially skewed results occur in New Haven, when better tests likely would have produced less disproportionate results?
The Court’s disparate-impact analysis rested on two provisions of Title VII: § 703(a)(2), which made it 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 race, color, religion, sex, or national origin”; and § 703(h), which permitted employers “to act upon the results of any professionally developed ability test provided that such test, its administration or action upon the results is not designed, intended or used to discriminate because of race, color, religion, sex or national origin.” Griggs v. Duke Power Co., 401 U. S. 424, 426, n. 1 (1971) (quoting 78 Stat. 255, 42 U. S. C. §2000e-2(a)(2), (h) (1964 *621ed.)). See also 401 U. S., at 433-436 (explaining that § 703(h) authorizes only tests that are “demonstrably a reasonable measure of job performance”).
See also Dothard v. Rawlinson, 433 U. S. 321, 332, n. 14 (1977) (“a discriminatory employment practice must be shown to be necessary to safe and efficient job performance to survive a Title VII challenge”); Williams v. Colorado Springs, Colo., School Dist., 641 F. 2d 835, 840-841 (CA10 1981) (“The term ‘necessity’ connotes that the exclusionary practice must be shown to be of great importance to job performance.”); Kirby v. Colony Furniture Co., 613 F. 2d 696, 705, n. 6 (CA8 1980) (“the proper standard for determining whether ‘business necessity’ justifies a practice which has a racially discriminatory result is not whether it is justified by routine business considerations but whether there is a compelling need for the employer to maintain that practice and whether the employer can prove there is no alternative to the challenged practice”); Pettway v. American Cast Iron Pipe Co., 494 F. 2d 211, 244, n. 87 (CA5 1974) (“this doctrine of business necessity ... connotes an irresistible demand” (inter*623nal quotation marks omitted)); United States v. Bethlehem, Steel Corp., 446 F. 2d 652, 662 (CA2 1971) (an exclusionary practice “must not only directly foster safety and efficiency of a plant, but also be essential to those goals”); Robinson v. Lorillard Corp., 444 F. 2d 791, 798 (CA4 1971) (“The test is whether there exists an overriding legitimate business purpose such that the practice is necessary to the safe and efficient operation of the business.”).
See, e.g., Nash v. Jacksonville, 837 F. 2d 1534 (CA11 1988), vacated, 490 U. S. 1103 (1989), opinion reinstated, 905 F. 2d 355 (1990); Vulcan Pioneers, Inc. v. New Jersey Dept. of Civil Serv., 832 F. 2d 811 (CA3 1987); Guardians Assn. of N. Y. City Police Dept. v. Civil Serv. Comm’n, 630 F. 2d 79 (CA2 1980); Ensley Branch of NAACP v. Seibels, 616 F. 2d 812 (CA5 1980); Firefighters Inst. for Racial Equality v. St. Louis, 616 F. 2d 350 (CA8 1980); Boston Chapter, NAACP, Inc. v. Beecher, 504 F. 2d 1017 (CA1 1974).
What was the “business necessity” for the tests New Haven used? How could one justify, e. g., the 60/40 written/oral ratio, see supra, at 611-612, 614-615, under that standard? Neither the Court nor the concurring opinions attempt to defend the ratio.
Even in Title VII cases involving race-conscious (or gender-conscious) affirmative-action plans, the Court has never proposed a strong-basis-inevidenee standard. In Johnson v. Transportation Agency, Santa Clara Cty., 480 U. S. 616 (1987), the Court simply examined the municipal employer’s action for reasonableness: “Given the obvious imbalance in the Skilled Craft category, and given the Agency’s commitment to eliminating such imbalances, it was plainly not unreasonable for the Agency ... to consider as one factor the sex of [applicants] in making its decision.” Id., at 637. See also Firefighters v. Cleveland,, 478 U. S. 501, 516 (1986) (“Title VII permits employers and unions voluntarily to make use of reasonable race-conscious affirmative action.”).
Notably, prior decisions applying a strong-basis-in-evidence standard have not imposed a burden as heavy as the one the Court imposes today. In Croson, the Court found no strong basis in evidence because the city *630had offered “nothing approaching a prima facie case.” Richmond v. J. A. Croson Co., 488 U. S. 469, 500 (1989). The Court did not suggest that anything beyond a prima facie ease would have been required. In the context of race-based electoral districting, the Court has indicated that a “strong basis” exists when the “threshold conditions” for liability are present. Bush v. Vera, 517 U. S. 952, 978 (1996) (plurality opinion).
Infecting the Court’s entire analysis is its insistence that the City rejected the test results “in sole reliance upon race-based statistics.” Ante, at 584. See also ante, at 580, 587. But as the part of the story the Court leaves out, see supra, at 609-618, so plainly shows — the long history of rank discrimination against African-Americans in the firefighting profession, the multiple flaws in New Haven’s test for promotions — “sole reliance” on statistics certainly is not descriptive of the CSB’s decision.
The Court’s refusal to remand for further proceedings also deprives respondents of an opportunity to invoke 42 U. S. C. §2000e-12(b) as a shield to liability. Section 2000e-12(b) provides:
“In any action or proceeding based on any alleged unlawful employment practice, no person shall be subject to any liability or punishment for or on account of (1) the commission by such person of an unlawful employment practice if he pleads and proves that the act or omission complained of was in good faith, in conformity with, and in reliance on any written interpretation or opinion of the [EEOC] .... Such a defense, if established, shall be a bar to the action or proceeding, notwithstanding that (A) after such act or omission, such interpretation or opinion is modified or rescinded or is determined by judicial authority to be invalid or of no legal effect....”
Specifically, given the chance, respondents might have called attention to the EEOC guidelines set out in 29 CFR §§ 1608.3 and 1608.4 (2008). The guidelines recognize that employers may “take affirmative action based on an analysis which reveals facts constituting actual or potential adverse impact.” § 1608.3(a). If “affirmative action” is in order, so is the lesser step of discarding a dubious selection device.
The lower courts focused on respondents’ “intent” rather than on whether respondents in fact had good cause to act. See 554 F. Supp. 2d 142, 157 (Conn. 2006). Ordinarily, a remand for fresh consideration would be in order. But the Court has seen fit to preclude further proceedings. I therefore explain why, if final adjudication by this Court is indeed appropriate, New Haven should be the prevailing party.
This alone would have posed a substantial problem for New Haven in a disparate-impact suit, particularly in light of the disparate results the City’s scheme had produced in the past. See supra, at 614. Under the *633Uniform Guidelines on Employee Selection Procedures (Uniform Guidelines), employers must conduct “an investigation of suitable alternative selection procedures.” 29 CFR § 1607.3(B). See also Officers for Justice v. Civil Serv. Comm’n, 979 F. 2d 721, 728 (CA9 1992) (“before utilizing a procedure that has an adverse impact on minorities, the City has an obligation pursuant to the Uniform Guidelines to explore alternative procedures and to implement them if they have less adverse impact and are substantially equally valid”). It is no answer to “presume” that the two-decades-old 60/40 formula was adopted for a “rational reason” because it “was the result of a union-negotiated collective-bargaining agreement.” Cf. ante, at 589. That the parties may have been “rational” says nothing about whether their agreed-upon selection process was consistent with business necessity. It is not at all unusual for agreements negotiated between employers and unions to run afoul of Title VII. See, e. g., Peters v. Missouri-Pacific R. Co., 483 F. 2d 490, 497 (CA5 1973) (an employment practice “is not shielded [from the requirements of Title VII] by the facts that it is the product of collective bargaining and meets the standards of fair representation”).
See also Nash, 837 F. 2d, at 1538 (“the examination did not test the one aspect of job performance that differentiated the job of firefighter engineer from fire lieutenant (combat): supervisory skills”); Firefighters Inst. for Racial Equality v. St. Louis, 549 F. 2d 506, 512 (CA8 1977) (“there is no good pen and paper test for evaluating supervisory skills”); Boston Chapter, NAACP, 504 F. 2d, at 1023 (“[T]here is a difference between memorizing . . . fire fighting terminology and being a good fire fighter. If the Boston Red Sox recruited players on the basis of their knowledge of baseball history and vocabulary, the team might acquire [players] who could not bat, pitch or catch.”).
Cf. Gillespie v. Wisconsin, 771 F. 2d 1035, 1043 (CA7 1985) (courts must evaluate “the degree to which the nature of the examination procedure approximates the job conditions”). In addition to “content validity,” the Uniform Guidelines discuss “construct validity” and “criterion validity” as means by which an employer might establish the reliability of a selection method. See 29 CFR § 1607.14(B)-(D). Content validity, however, is the only type of validity addressed by the parties and “the only feasible type of validation in these circumstances.” Brief for Industrial-Organizational Psychologists as Amicus Curiae 7, n. 2 (hereinafter 1-0 Psychologists Brief).
See also G. Thornton & D. Rupp, Assessment Centers in Human Resource Management 15 (2006) (“Assessment centers predict future success, do not cause adverse impact, and are seen as fair by participants.”); W. Cascio & H. Aguinis, Applied Psychology in Human Resource Management 372 (6th ed. 2005) (“research has demonstrated that adverse impact is less of a problem in an [assessment center] as compared to an aptitude test”). Cf. Firefighters Inst. for Racial Equality, 549 F. 2d, at 513 (recommending assessment centers as an alternative to written exams).
Finding the evidence concerning these alternatives insufficiently developed to “create a genuine issue of fact,” ante, at 591, the Court effectively confirms that an employer cannot prevail under its strong-basis-*636in-evidence standard unless the employer decisively proves a disparate-impact violation against itself. The Court’s specific arguments are unavailing. First, the Court suggests, changing the oral/written weighting may have violated Title VII’s prohibition on altering test scores. Ante, at 590. No one is arguing, however, that the results of the exams given should have been altered. Rather, the argument is that the City could have availed itself of a better option when it initially decided what selection process to use. Second, with respect to assessment centers, the Court identifies “statements to the CSB indicat[ing] that the Department could not have used [them] for the 2003 examinations.” Ante, at 591. The Court comes up with only a single statement on this subject — an offhand remark made by petitioner Ricci, who hardly qualifies as an expert in testing methods. See ante, at 574. Given the large number of municipalities that regularly use assessment centers, it is impossible to fathom why the City, with proper planning, could not have done so as well.
The I-O Psychologists Brief identifies still other, more technical flaws in the exams that may well have precluded the City from prevailing in a *638disparate-impaet suit. Notably, the exams were never shown to be suitably precise to allow strict rank ordering of candidates. A difference of one or two points on a multiple-choice exam should not be decisive of an applicant’s promotion chances if that difference bears little relationship to the applicant’s qualifications for the job. Relatedly, it appears that the fine between a passing and failing score did not accurately differentiate between qualified .and unqualified candidates. A number of fire-officer promotional exams have been invalidated on these bases. See, e. g., Guardians Assn., 630 F. 2d, at 105 (“When a cutoff score unrelated to job performance produces disparate racial results, Title VII is violated.”); Vulcan Pioneers, Inc. v. New Jersey Dept. of Civil Serv., 625 F. Supp. 527, 538 (NJ 1985) (“[T]he tests here at issue are not appropriate for ranking candidates.”).
Some of petitioners’ so-called facts find little support in the record, and many others can scarcely be deemed material. Petitioners allege, for example, that City officials prevented New Haven’s fire chief and assistant chief from sharing their views about the exams with the CSB. App. to Pet. for Cert. in No. 07-1428, p. 228a. None of the materials petitioners dte, however, “suggests” that this proposition is accurate. Cf. ante, at 600. In her deposition testimony, City official Karen Dubois-Walton specifically denied that she or her colleagues directed the chief and assistant chief not to appear. App. to Pet. for Cert. in No. 07-1428, p. 850a. Moreover, contrary to the insinuations of petitioners and Justice Alito, the statements made by City officials before the CSB did not emphasize allegations of cheating by test takers. Cf. ante, at 602-603. In her deposition, Dubois-Walton acknowledged sharing the cheating allegations not with the CSB, but with a different City commission. App. to Pet. for Cert. in No. 07-1428, p. 837a. Justice Alito also reports that the City’s attorney advised the mayor’s team that the way to convince the CSB not to certify was “to focus on something other than 'a big discussion re: adverse impact’ law.” Ante, at 603 (quoting App. to Pet. for Cert. in No. 07-1428, *640p. 458a). This is a misleading abbreviation of the attorney’s advice. Focusing on the exams’ defects and on disparate-impact law is precisely what he recommended. See id., at 458a-459a.
City officials, Justice Alito reports, sent Hornick newspaper accounts and other material about the exams prior to his testimony. Ante, at 603. Some of these materials, Justice Alito intimates, may have given Hornick an inaccurate portrait of the exams. But Hornick’s testimony before the CSB, viewed in full, indicates that Hornick had an accurate understanding of the exam process. Much of Horniek’s analysis focused on the 60/40 weighting of the written and oral exams, something that neither the Court nor the concurrences even attempt to defend. It is, moreover, entirely misleading to say that the City later hired union-proposed Hornick as a “rewar[d]” for his testimony. Cf. ibid.
To be dear, the board of fibre commissioners on which Kimber served is an entity separate from the CSB. Kimber was not a member of the CSB. Kimber, Justice Auto states, requested a private meeting with the CSB. Ante, at 601. There is not a shred of evidence that a private meeting with Kimber or anyone else took place.
Justice Auto points to evidence that the mayor had decided not to make promotions based on the exams even if the CSB voted to certify the results, going so far as to prepare a press release to that effect. Ante, at 604. If anything, this evidence reinforces the conclusion that the CSB — which made the noneertification decision — remained independent and above the political fray. The mayor and his staff needed a contingency plan precisely because they did not control the CSB.
The District Court, Justice Alito writes, “all but conceded that a jury could find that the City’s asserted justification was pretextual” by “admitting] that 'a jury could rationally infer that city officials worked behind the scenes to sabotage the promotional examinations because they knew that, were the exams certified, the Mayor would incur .the wrath of *644[Rev. Boise] Kimber and other influential leaders of New Haven’s African-American community.’” Ante, at 598, 608 (quoting 554 F. Supp. 2d, at 162). The District Court drew the quoted passage from petitioners’ lower court brief, and used it in reference to a First Amendment claim not before this Court. In any event, it is not apparent why these alleged political maneuvers suggest an intent to discriminate against petitioners. That City officials may have wanted to please political supporters is entirely consistent with their stated desire to avoid a disparate-impact violation. Cf. Ashcroft v. Iqbal, 556 U. S. 662, 682 (2009) (allegations that senior Government officials condoned the arrest and detention of thousands of Arab Muslim men following the September 11 attacks failed to establish even a “plausible inference” of unlawful discrimination sufficient to survive a motion to dismiss).