Opinion for the court filed by Circuit Judge MIKVA.
Opinion concurring in part and dissenting in part filed by Circuit Judge BUCKLEY.
MIKVA, Circuit Judge:
This attorney’s fee dispute presents the question whether the district court abused its discretion in awarding an enhancement of attorney’s fees and costs based on the risk of nonpayment (i.e., the contingent nature of the case) and the quality of representation. Following Pennsylvania v. Delaware Valley Citizens’ Council for Clean Air, 478 U.S. 546, 106 S.Ct. 3088, 92 L.Ed.2d 439 (1986) (“Delaware Valley I”), and Pennsylvania v. Delaware Valley Citizens’ Council for Clean Air, 483 U.S. 711, 107 S.Ct. 3078, 97 L.Ed.2d 585 (1987) (“Delaware Valley II”), we affirm the district court’s award.
I. Inteoduction
This case involves the fees and costs of counsel for a class of black employees at the Government Printing Office (“GPO”), whose legal odyssey has stretched over the better part of two decades. In 1973, the employees (“plaintiffs”) filed an action alleging racial discrimination in hiring, training, and promotion practices, in violation of Title VII of the Civil Rights Act of 1964, which had been extended in 1973 to cover federal workers, see 42 U.S.C. § 2000e-16. In 1977, the district court granted plaintiffs’ motion for summary judgment on all claims of liability under Title VII, see McKenzie v. McCormick, 425 F.Supp. 137, 142 (D.D.C.1977), and four years later, the district court issued its remedial decree, see McKenzie v. Saylor, 508 F.Supp. 641, 647-59 (D.D.C.1981). On appeal, this court affirmed the district court in large part, see McKenzie v. Sawyer, 684 F.2d 62, 80 (D.C.Cir.1982).
Although the suit was both a class and an individual action, only the claims of counsel for the class are involved in this appeal. Class counsel include the Institute for Public Representation (“Institute”), [332]*332part of the clinical education program of the Georgetown University Law Center; the Washington Lawyers’ Committee for Civil Rights Under Law (“Lawyers’ Committee”), a non-profit, tax-exempt organization affiliated with the National Lawyers Committee for Civil Rights; and the law firm of Hogan & Hartson. Class counsel (collectively “applicants”) first filed a petition requesting attorney’s fees and costs in April 1981, pursuant to 42 U.S.C. §§ 2000e-5(k), 2000e-16(d). Applicants were granted an interim award covering the period from the outset of the case until January 30, 1981, the date of the final relief order. See McKenzie v. Kennickell, 645 F.Supp. 437, 441 (D.D.C.1986). After some considerable delay, the government eventually complied with the district court’s order, see McKenzie v. Kennickell, 669 F.Supp. 529, 530-31, 535 (D.D.C.1987).
Following the resolution of the litigation on the merits, applicants again petitioned the district court for an interim award of attorney’s fees (for services since 1981) pending a final determination of fees. On August 10, 1987, the district court ordered the government to identify an amount that in its opinion represented the “irreducible minimum lodestar fee” to which applicants were entitled, see 669 F.Supp. at 531. Under this order, the government ultimately paid applicants some $200,000, see 669 F.Supp. at 536.
On April 8, 1988, the district court approved a final stipulation in which the government agreed to a lodestar fee of $740,000 that covered the total fee claims of counsel for the individual plaintiffs and counsel handling all fee claims, and the lodestar claims of counsel for the class. This sum was paid on April 28, 1988. See McKenzie v. Kennickell, 684 F.Supp. 1097, 1098 (D.D.C.1988).
On April 18, 1988, the district judge granted counsel for the class two enhancements of the lodestar award: a 50 percent enhancement to reflect the contingent nature of the claim and the resulting risk of nonpayment, and a 25 percent enhancement on the basis of quality of representation. See McKenzie, 684 F.Supp. at 1099. The government appeals these enhancements.
II. Discussion
A. The Contingency Enhancement
In Delaware Valley II, the Supreme Court specifically addressed the legal standard for contingency enhancements under fee-shifting statutes. A majority of the Court reversed a 100 percent risk enhancement before it, and a plurality of four contended that contingency enhancements “should be reserved for exceptional cases.” 107 S.Ct. at 3088. But Justice O’Connor, in a controlling opinion, concluded that “Congress did not intend to foreclose consideration of contingency in setting a reasonable fee under fee-shifting provisions,” 107 S.Ct. at 3089 (O’Connor, J., concurring in part and concurring in the judgment). Four Justices in dissent agreed, see 107 S.Ct. at 3091 (Blackmun, J., dissenting, joined by Brennan, Marshall, and Stevens, JJ.).
Justice O’Connor’s opinion in Delaware Valley II is currently the effective legal standard for contingency enhancements, see Weisberg v. U.S. Dep’t of Justice, 848 F.2d 1265, 1272 (D.C.Cir.1988); Thompson v. Kennickell, 836 F.2d 616, 621 (D.C.Cir.1988). Her test was twofold. First, she urged lower courts to view contingency cases “as a class” and “treat a determination of how a particular market compensates for contingency as controlling future cases involving the same market.” 107 S.Ct. at 3090. Justice O’Connor recommended that “courts strive for consistency from one fee determination to the next.” Id. She emphasized that “at all times a fee applicant bears the burden of proving the degree to which the relevant market compensates for contingency.” Id. Second, Justice O’Connor agreed with the plurality that “no enhancement for risk is appropriate unless the applicant can establish that without an adjustment for risk the prevailing party ‘would have faced substantial difficulties in finding counsel in the local or other relevant market.’ ” Id. at 3091 (quoting plurality opinion, 107 S.Ct. at 3089).
Justice O’Connor’s opinion must be understood with reference to the overall pro[333]*333cess by which attorney’s fees are awarded, because she recognized that contingency is but one factor that a district court may consider in setting a “reasonable fee,” 107 S.Ct. at 3089. As the Supreme Court recently reaffirmed, “reasonableness” is the overarching standard by which a fee award is to be measured, and it is to be determined by a district court “in light of all the circumstances” of a case. Blanchard v. Bergeron, — U.S. -, 109 S.Ct. 939, 944, 103 L.Ed.2d 67 (1989). A district court judge has broad discretion to examine the relevant factors in each individual case; “[i]t is central to the awarding of attorney’s fees under § 1988 that the district court judge, in his or her good judgment, make the assessment of what is a reasonable fee under the circumstances of the case.” Blanchard, 109 S.Ct. at 946. This means that the size of the contingency enhancement may vary from case to case; a district court may often find it useful to set the contingency enhancement last, after other enhancements have been made, so that the total fees awarded do not exceed what is “reasonable.”
We note that in some areas Congress has seen fit to prescribe a particular percentage of the recovery as a cap on recoverable fees, see, e.g., 42 U.S.C. § 406(b)(1) (establishing 25 percent of recovery as ceiling in social security disability cases). This parallels the market practice whereby fees are often set as a percentage (say, 33 percent) of the recovery. Such a rule cannot be rigidly applied under fee-shifting statutes, because of the frequent inapplicability of “the private market model of contingency compensation,” 107 S.Ct. at 3090 (O’Con-nor, J., concurring in part and concurring in the judgment), and because under most statutes the opponent rather than the client pays the fee. See Rodriquez v. Bowen, 865 F.2d 739 (6th Cir.1989) (en banc). Still, a district court often may wish to compare the fee award to the overall recovery, in order to ensure that the market practice is not exceeded and the fee remains reasonable.
With these principles in mind, we turn to the issues presented by the case sub judi-ce.
1. Whether Applicants Are Eligible for an Enhancement
At the outset, the government maintains that the Institute and the Lawyers’ Committee are not eligible for enhancements for the risk of nonpayment. The government contends that because these organizations do not accept fee-paying clients, they did not incur any opportunity cost in representing the plaintiffs in this action. They did not, in other words, fore-go any work for which they would have been compensated on an hourly basis. Furthermore, neither accepted the case on the expectation of a contingency enhancement; indeed, their raison d’etre is to assist litigants who cannot afford to pay fees. See New York Ass’n for Retarded Children v. Carey, 711 F.2d 1136, 1154 (2d Cir.1983) (holding unreasonable the addition of contingency bonuses to fee awards to non-profit law offices).
The Supreme Court, however, has stressed that the computation of fees does not vary according to the identity of the recipient. “Congress did not intend the calculation of the awards to vary depending on whether the plaintiff was represented by private counsel or by a nonprofit legal services organization.” Blum v. Stenson, 465 U.S. 886, 894, 104 S.Ct. 1541, 1547, 79 L.Ed.2d 891 (1984); see also id. at 904, 104 S.Ct. at 1551 (noting that risk enhancements are available for “nonprofit legal service organizations and private attorneys” on an equal basis) (Brennan, J., concurring). Indeed, as the Supreme Court recently opined, the fact “[t]hat a nonprofit legal services organization may contractually have agreed not to charge any fee of a civil rights plaintiff does not preclude the award of a reasonable fee to a prevailing party in a § 1983 action, calculated in the usual way.” Blanchard v. Bergeron, 109 S.Ct. 939, 945 (1989) (emphasis in original). If this were not so, the Delaware II Court need not have reached the question of the appropriate standards for contingency enhancements generally, because it could have found that non-profit law firms were ineligible for such enhancements. See 107 [334]*334S.Ct. at 3088, n. 10 (noting that the attorneys for Delaware Valley were part of a non-profit, tax-exempt law corporation, but expressly declining to pass on argument— not aired in the lower courts — that such a firm, by its very nature, is ineligible for a risk-of-not-prevailing enhancement); 107 S.Ct. at 3096 (Blackmun, J., dissenting) (rejecting the argument that non-profits are ineligible for contingency enhancements as an “attempt to do indirectly what the Court refused to do directly in Blum”); see also Blanchard, 109 S.Ct. at 945.
In addition, this circuit has rejected the notion that the identity of counsel should play any role in the determination of fees. See Save Our Cumberland Mountains, Inc. v. Hodel, 857 F.2d 1516, 1521-24 (D.C.Cir.1988) (en banc). Far from constituting “economic relief” or a “windfall” to nonprofit firms, the equal availability of contingency enhancements creates a level playing field. If enhancements were available to for-profit firms, but not to non-profits, then over time the chronic imbalance would encourage legal resources to move to the for-profit sector where the risk of nonpayment could be fully compensated. This would inhibit the Institute and Lawyers’ Committee from bringing contingent Title VII cases and subvert the congressional intent that fee-shifting statutes should benefit non-profits. See New York Gaslight Club, Inc. v. Carey, 447 U.S. 54, 70 n. 9, 100 S.Ct. 2024, 2034 n. 9, 64 L.Ed.2d 723 (1980).
We conclude that applicants are eligible for a contingency enhancement.
2. The First Prong: Whether “The Relevant Market” Adds a Premium for Contingency
We proceed to the first prong of Justice O’Connor’s test: whether the relevant legal market adds a premium for contingency. The term “relevant market” is a source of disagreement between the parties. Applicants urge that it be defined as including all contingent cases in the contemporary Washington, D.C., legal market; the government contends that the phrase should be interpreted to encompass only Title VII and related cases at the time plaintiffs brought their suit.
We hold that the “relevant market” in this case is composed of all contingency claims in the District of Columbia, particularly other types of complex federal litigation, at the time the case was undertaken by counsel (rather than at the time that fees were awarded). Throughout her concurrence, Justice O’Connor referred to “the class” of contingency cases, without suggesting that the class should be restricted according to the subject matter of the plaintiff’s complaint. See, e.g., 107 S.Ct. at 3090. “It does not appear that Justice O’Connor * * * contemplated that the class of cases to be studied be anything less than all contingency cases in a given geographic market * * *.” Blum v. Witco Chemical Corp., 829 F.2d 367, 381 (3d Cir.1987); see also Blum v. Stenson, 465 U.S. 886, 890-91, 892 & n. 5, 104 S.Ct. 1541, 1544-45 & n. 5, 79 L.Ed.2d 891 (1984) (referring to “the prevailing market rate” for attorneys of comparable experience, skill, and reputation) (emphasis added). This definition of “relevant market” is consistent with congressional intent, see S.Rep. No. 1011, 94th Cong., 2d Sess. 6 (1976), U.S.Code Cong. & Admin.News 1976, p. 5913 (“It is intended that the amount of fees awarded under [section 1988] be governed by the same standards which prevail in other types of equally complex Federal litigation, such as antitrust cases[,] and not be reduced because the rights involved may be nonpecu-niary in nature.”).
We also hold that the “relevant market” is that which existed at the time the case was undertaken by counsel and the suit commenced, rather than the market existing at the time that fees were awarded. Congress has instructed that fees be set according to “what is traditional with attorneys compensated by a fee-paying client.” S.Rep. No. 1011, 94th Cong., 2d Sess. 6 (1976), U.S.Code Cong. & Admin.News 1976, p. 5913; see also H.R.Rep. No. 1558, 94th Cong., 2d Sess. 9 (1976). In private markets, the fee is negotiated up front at the beginning of a case. No lawyer would seek to enlarge a contingency premium, or create one where none existed before, if [335]*335the community practice changed during the course of the lawsuit. Similarly, we see no need to reward applicants in this case if, since the time they agreed to represent the class of GPO employees, fee practices in competitive markets have changed. We remind applicants that a “reasonable” fee is one that is “adequate to attract competent counsel, but * * * [that does] not produce windfalls to attorneys.” S.Rep. No. 1011, 94th Cong., 2d Sess. 6 (1976), U.S.Code Cong. & Admin.News 1976, p. 6913.
Focusing on the time when the case was commenced provides practitioners more predictability than would the contrary approach. A lawyer deciding whether to take on a case, on the basis of fees provided under a statute, can ascertain whether a contingency enhancement will be available by examining then-existing market practices. She need not be troubled by the possibility that those practices might later change (say, by making contingency enhancements unavailable) before the suit is completed and she applies for fees.
The government urges that “relevant market” be defined according to the antitrust notion of “market”: a group of goods as defined by the cross-elasticity of demand between a product and its substitutes. See Rothery Storage & Van Co. v. Atlas Van Lines, Inc., 792 F.2d 210, 218-19 (D.C.Cir.1986) (Bork, J.), cert. denied, 479 U.S. 1033, 107 S.Ct. 880, 93 L.Ed.2d 834 (1987). Thus, the government would have us define the relevant market in the case sub judice by looking at actions in the subject areas of law — age discrimination, civil rights, or commercial litigation — whose fees influence, and are influenced by, rates of compensation in the Title YII area. According to the government, a showing that personal injury or antitrust lawyers are usually compensated for contingent risk cannot justify a contingency enhancement in this case, unless applicants can demonstrate that those lawyers would accept a Title VII action against the government. If applicants can show no broader market as evidenced by high cross-elasticity of demand, then they must show that the Title VII market itself provides an enhancement for the risk of not prevailing. The government cites Fadhl v. City and County of San Francisco, 859 F.2d 649, 650 (9th Cir.1988) (per curiam) (defining the “relevant market” as “Title VII cases in San Francisco”).
We are unpersuaded by this argument. First, Title VII cases by themselves cannot qualify as a “relevant market” because “these types of cases are never compensated on a straight contingency and hence there is no relevant market for comparison.” Blum v. Witco Chemical Corp., 829 F.2d 367, 381 (3d Cir.1987). The same can be said for many environmental and other cases brought under fee-shifting statutes, which have no private market analogues. See Delaware II, 107 S.Ct. at 3090 (O’Con-nor, J., concurring in part and concurring in the judgment).
As a result, fee petitioners would be forced to focus entirely on the types of cases that show a high cross-elasticity of demand with respect to Title VII actions. Applicants would be required to conduct elaborate market surveys of attorneys in particular locales, showing the number of lawyers willing to accept Title VII cases at various hypothetical levels of fee awards. The parties would clash over the accuracy of the data, the appropriate cross-elasticities to be calculated, the existence of possible sub-markets, the extent of product differentiation within the legal market as a whole, and the significance of barriers to entry, to name but a few topics. We need only glance at contemporary antitrust litigation, with its momentous battles over the definition of the appropriate “market,” to see how much litigation would be created in practice by the government’s proposal. See P. Areeda & L. Kaplow, Antitrust Analysis ¶¶ 341-45 (4th ed. 1988).
The Supreme Court, of course, has inveighed against precisely this type of approach to fee-setting. “Fee litigation * * * is often protracted, complicated, and exhausting. There is little doubt that it should be simplified to the maximum extent possible.” Delaware Valley II, 107 S.Ct. at 3085 (plurality opinion); see also Hensley v. Eckerhart, 461 U.S. 424, 437, 103 S.Ct. 1933, 1941, 76 L.Ed.2d 40 (1983) (“A request for attorney’s fees should not result in a second major litigation”). The government’s proposal would result in ap[336]*336pendage litigation that would dwarf the principal actions.
For similar reasons we reject the government’s suggestion that the contingency premium should be measured by some sort of statistically valid market survey, rather than merely by affidavits produced by a fee petitioner. The type of scientifically controlled survey sought by the government again would invite a second litigation over fees. In addition, there is nothing in Supreme Court precedent or the legislative history of fee-shifting statutes that demands such an expensive, exhaustive, and stringent measure of proof. Justice O’Connor, in Delaware Valley II, merely required district courts to make specific “findings of fact,” 107 S.Ct. at 3091, indicating why an enhancement was proper. This was also the approach used in Blum v. Stenson, 465 U.S. 886, 898-99, 104 S.Ct. 1541, 1548-49, 79 L.Ed.2d 891 (1984).
We are satisfied that applicants in this case have met their burden. They submitted more than 20 affidavits from practitioners to bolster their claim that the enhancement should amount to at least 50 percent. The affidavits in fact supported a wide spectrum of enhancements, ranging from 5 percent to 33y3 percent and 300 percent. The district court found that these were adequate to justify an enhancement of 50 percent, see McKenzie, 684 F.Supp. at 1102-03, and we agree. While much of applicants’ evidence concerns current market practices rather than the market practices that existed in 1973 and 1974, when the case was undertaken by counsel, we find that applicants have shown that this contemporary evidence is equally applicable to the historic 1973-74 time frame. The district court found that “fee petitioners have demonstrated that the legal market required compensation for contingency during the 1970s. Several of the affiants discussed their firm[s’] policies in the 1970’s and stated that a premium for contingency was routinely required.” McKenzie, 684 F.Supp. at 1102. After examining the record, we agree with the district court. We note that applicants’ evidence must be evaluated in terms of the types of contingency arrangements existing at the time; the lodestar method of analysis was not announced until Lindy Bros. Builders, Inc. v. American Radiator & Standard Sanitary Corp., 487 F.2d 161, 167 (3d Cir.1973), and first made its appearance in this circuit in Copeland v. Marshall, 641 F.2d 880 (D.C.Cir.1980) {en banc), so that we would not expect to see any modern form of contingency enhancement reflected in applicants’ submissions.
The enhancement in this case is also consistent with Justice O’Connor’s recommendation that the lower courts in each relevant market area should determine the appropriate enhancement to be used in entire classes of contingent cases. “District Courts and Courts of Appeals should treat a determination of how a particular market compensates for contingency as controlling future cases involving the same market.” 107 S.Ct. at 3090 (O’Connor, J., concurring in part and concurring in the judgment). We note that 50 percent is, if anything, below the Title VII contingency enhancements typically awarded in this circuit. See Thompson v. Kennickell, Civ. No. 74-1101, Mem. op. at 2, 16-17 (D.D.C. Mar. 30, 1989) (granting a 100 percent contingency enhancement in a Title VII and Equal Pay Act class action); Palmer v. Shultz, 679 F.Supp. 68, 74-76 (D.D.C.1988), appeal dism’d, No. 88-5108 (D.C.Cir.1988) (awarding a 50 percent risk enhancement in a partially contingent Title VII case in which the client agreed to pay half the usual contingent fee, win or lose); King v. Palmer, Civ. No. 83-1980, Mem. op. at 4-5, 1988 WL 104970 (D.D.C. Sept. 20, 1988) (awarding a 50 percent enhancement in a partially contingent Title VII case and noting that “[a] simple and fair application of these principles would establish 100 percent enhancement for contingent fees, and 50 percent for partially contingent fees”). We find that the award of 50 percent in this case for a wholly contingent risk was not excessive.
3. The Second Prong: Whether, in the Absence of an Enhancement for Risk, Plaintiffs Would Have Faced Substantial Difficulty in Finding Counsel
Justice O’Connor also stated that “no enhancement of risk is appropriate un[337]*337less the applicant can establish that without an adjustment for risk the prevailing party ‘would have faced substantial difficulties in finding counsel in the local or other relevant market.’ ” Delaware Valley II, 107 S.Ct. at 3091 (O’Connor, J., concurring in part and concurring in the judgment) (quoting plurality opinion). We hold that applicants are required to demonstrate that absent a contingency enhancement, plaintiffs would have encountered substantial difficulty in finding counsel in 1973 and 1974, when they commenced their lawsuit. This is necessarily a counterfactual inquiry; applicants are not required to show that plaintiffs actually did face difficulty.
A standard of actual difficulty, urged by the government in this case, would have perverse effects in practice. It would penalize plaintiffs who were lucky enough to stumble across on their first try the one-in-a-hundred lawyer who was willing to take their ease. It would also discourage referral services such as the Lawyers’ Committee that make it easier for litigants to find legal representation. Cf. Texas State Teachers Ass’n v. Garland Independent School District, — U.S. -, 109 S.Ct. 1486, 1494, 103 L.Ed.2d 866 (1989) (O’Con-nor, J.) (noting that the Civil Rights Attorney’s Fees Awards Act of 1976 was meant to “promote” service by counsel in the “private attorney general” role). Finally, it would lead to a charade in which clients seeking representation under fee-shifting statutes would be steered to several attorneys whose pre-arranged role it would be to “refuse” the case, knowing that such refusals were necessary to permit the eventual award of fees. For these reasons, we conclude that applicants need not show that plaintiffs actually experienced difficulty in obtaining representation. They must only establish, according to the test formulated by Justice O’Connor, that plaintiffs would have faced “substantial difficulties” in the absence of a contingency enhancement.
The district court found that today “there exists a general dearth of counsel willing to accept [Title VII] suits in the absence of [a contingency] enhancement,” McKenzie, 684 F.Supp. at 1103, and con-eluded that applicants had shown that but for the availability of a contingency enhancement, plaintiffs would have encountered substantial difficulty in securing representation. We repeat that the relevant time frame is the period in which the case was undertaken by counsel and the suit commenced, rather than the time the fees were awarded.
We believe that applicants have met their burden in this case. Although much of their evidence pertains to conditions in the current legal market, they have demonstrated that in 1973 and 1974 counsel in Title VII actions expected some form of enhancement for the contingent character of a case. See Stipulation Concerning the Testimony of the Hon. John Ferren, Joint Appendix (“J.A.”) at 67, 68 (“private attorneys general” during the 1972-73 period expected to be compensated “by awards of reasonable attorneys fees”); Declaration of Julius Chambers, J.A. at 156 (“Title VII cases present tremendous risks [to practitioners] irrespective of their ultimate outcomes. This was even more true in 1972 and 1973 than now, when many issues under Title VII were still largely unsettled.”); Declaration of Bradley G. McDonald, J.A. at 113 (noting he “anticipated” a contingency enhancement when he agreed to serve as counsel in a Title VII class action in 1972).
The government raises two objections to the evidence adduced and the findings predicated on such evidence. First, it challenges the sufficiency of the evidence introduced by applicants to demonstrate the need for an enhancement. The government maintains that the affidavits, which by their nature are based on opinion and speculation, are not scientific enough to establish that “adjustment for contingency was a crucial factor in plaintiffs’ ability to obtain counsel.” Jenkins v. State of Missouri, 838 F.2d 260, 268 (8th Cir.), cert. granted, — U.S. -, 109 S.Ct. 218, 102 L.Ed.2d 209 (1988). This contention is similar to the adequacy of the evidence argument made by the government under the first prong of Justice O’Connor’s test, and we reject it in this context as well. In [338]*338short, we find no reason to require elaborate statistical proof by applicants. Evidence in the form of affidavits from practitioners is acceptable, and the conclusions of fact of the district court in this case are not clearly erroneous, see Weisberg, 848 F.2d at 1268.
Second, the government argues that a contingency fee enhancement was unnecessary in this case to attract the particular counsel who represented plaintiffs. Both the Institute and the Lawyers’ Committee specialize in representing clients who cannot afford to pay, and the head of the Hogan & Hartson Community Services Department stated that in his opinion his firm would not have refused to represent the McKenzie plaintiffs even if it believed that no enhancement to the lodestar fee would be available.
We reject the notion that applicants must demonstrate that they, or any other particular counsel who represented plaintiffs, were specifically attracted by the possibility of a contingency enhancement. Such a standard would contravene the settled doctrine discussed earlier that fees must be awarded regardless of the identity of the fee petitioner. To deny a contingency enhancement on the ground that a public interest firm took the case without an expectation of a risk premium, or without the expectation of any fees at all, would run counter to Blum v. Stenson, 465 U.S. 886, 894, 104 S.Ct. 1541, 1546, 79 L.Ed.2d 891 (1984). In addition, the government’s approach would invite a second litigation over fees, as the parties sought to establish the subjective expectations of plaintiffs’ counsel. This prospect would encourage a charade similar to the one discussed above: public interest lawyers would accept a case only after announcing loudly that they were doing so on the assumption of a contingency enhancement. Justice O’Connor’s opinion instructs us to adopt a class-wide view of contingent cases; if the unavailability of risk enhancements would have caused plaintiffs to have experienced “substantial difficulty” in locating counsel, then, notwithstanding the particular circumstances of their case, such an enhancement may be granted. We agree with the district court that, absent an enhancement, the plaintiffs in this case would have encountered such difficulty.
B. The Quality Enhancement
The district court also awarded a 25 percent enhancement on the basis of “the quality of representation and exceptional results.” McKenzie, 684 F.Supp. at 1105. The district court acknowledged that enhancements made on the basis of superior quality of representation are available only in “rare” and “exceptional” cases, 684 F.Supp. at 1105 (quoting Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098), but contended that “[i]f the Supreme Court’s caveat authorizing enhancements in rare and exceptional cases is to be more than a theoretical possibility, then this proceeding is one of such cases, deserving of an enhancement.” 684 F.Supp. at 1105. We agree, and uphold the district court’s award.
The Supreme Court has taught that “ ‘the special skill and experience of counsel,’ the ‘quality of representation,’ and the ‘results obtained’ from the litigation are presumably fully reflected in the lodestar amount, and thus cannot serve as independent bases for increasing the basic fee award.” Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098 (quoting Blum, 465 U.S. at 898-900, 104 S.Ct. at 1548-49). Enhancements can only be made after “ ‘specific evidence’ on the record and detailed findings by the lower courts.” Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098 (quoting Blum, 465 U.S. at 899, 104 S.Ct. at 1549); see also Norman v. Housing Authority of City of Montgomery, 836 F.2d 1292, 1302 (11th Cir.1988).
In this case, the district court made specific findings that the attorneys for whom a low rate was charged in the lodestar in fact performed in a manner worthy of much higher rates, see McKenzie, 684 F.Supp. at 1107. It noted that a good deal of the work on the case was done by two junior associates who “performed at levels well beyond the standards expected of attorneys with such limited experience.” 684 F.Supp. at 1107. The two remained active [339]*339in the litigation over a period of 15 years, an “extremely rare” occurrence. Id. The district court, which was involved with the litigation from the beginning, was in a good position to evaluate the quality of the lawyering performed. “The court that tries a case is best qualified to make an appropriate allowance of attorney fees,” Craig v. Secretary, Department of Health and Human Services, 864 F.2d 324, 328 (4th Cir.1989), especially with respect to a quality enhancement.
The government contends that the district court erroneously relied upon the “exceptional results” obtained by the applicants, in awarding them an enhancement. The results, according to the government, are an improper factor for the district court to consider because they presumably are already included in the lodestar. See Delaware Valley I, 478 U.S. at 566, 106 S.Ct. at 3098 (rejecting enhancement based on “outstanding result” obtained).
We find that the district court did not err in this case. While it would be error to award an enhancement solely on the basis of “exceptional results,” it is not error to consider such results as a threshold requirement to be met before an enhancement for quality can be awarded. Indeed, it would be peculiar to reward a high quality of representation which led to only mediocre results. We find that the district court in fact followed this reasoning. It recognized that “our Court of Appeals has instructed that remarkable success, alone, is insufficient to warrant an enhancement,” McKenzie, 684 F.Supp. at 1106 and maintained merely that “the degree of success cannot be ignored,” id., and “a review of the exceptional results is integral to an analysis of the quality of representation.” Id. The district court thus properly considered the results obtained as a necessary, but not a sufficient, factor in awarding an enhancement.
III. Conclusion
Fee applicants must establish that at the time counsel undertook the case, contingent claims — especially in other types of complex federal litigation — in the relevant geographic area (here, Washington, D.C.) received a premium for the risk of non-payment; and that in the absence of a contingency enhancement plaintiffs would have experienced substantial difficulty in obtaining representation, irrespective of whether plaintiffs actually did encounter difficulty in locating counsel, or whether the particular counsel ultimately secured by plaintiffs entered the case with the subjective expectation of a contingency enhancement. These showings can be made by introducing affidavits from knowledgeable practitioners and other parties. We reiterate that the risk of not prevailing is one factor in determining the reasonableness of a fee award, and that a district court has the obligation to ensure that the overall award does not exceed a reasonable amount.
We also find that the district court’s award of an enhancement based on quality of representation was accompanied by sufficiently specific findings of fact. The district court properly interpreted “exceptional results” as a threshold to be met before a quality enhancement could be considered and awarded.
We affirm the district court’s award of enhancements for both contingency and quality of representation.
It is so ordered.