BALDOCK, Circuit Judge.
The Civil Rights Attorney’s Fees Awards Act of 1976 provides that, in a civil rights action, “the court, in its discretion, may allow the prevailing party, other than the United States, a reasonable attorney’s fee as part of the costs.” 42 U.S.C. § 1988. The presumptively “reasonable attorney’s fee” contemplated by § 1988 is the “ ‘product of reasonable hours times a reasonable rate.’ ” Blum v. Stenson, 465 U.S. 886, 897, 104 S.Ct. 1541, 1548, 79 L.Ed.2d 891 (1984) (quoting Hensley v. Eckerhart, 461 U.S. 424, 434, 103 S.Ct. 1933, 1940, 76 L.Ed.2d 40 (1983)). See also Pennsylvania v. Delaware Valley Citizens’ Council (Delaware Valley I), 478 U.S. 546, 565, 106 S.Ct. 3088, 3098, 92 L.Ed.2d 439 (1986). In the parlance of attorney’s fee litigation, this figure is known as the “lodestar.” See Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098. The narrow issue presented [1354]*1354by this case is if and when “a reasonable attorney’s fee” should include an enhancement to the lodestar for the risk of losing and not being paid.1
Plaintiff-appellant Homeward Bound, Inc. (“HBI”) is an organization of parents of developmentally disabled children who resided at the Hissom Memorial Center (“HMC”), one of the three largest state institutions in Oklahoma for the developmentally disabled. HBI sought to reform the state’s treatment of the residents through litigation. However, HBI had difficulty securing the services of an attorney. Eastern Oklahoma Legal Aid, the American Civil Liberties Union of Oklahoma, and Protection and Advocacy all indicated that they did not have the resources to represent HBI in the large institutional reform litigation presented by the situation at HMC. Several private attorneys, both local and out of state, echoed similar concerns and expressed their unwillingness to represent HBI without a commitment of substantial resources. After nearly a two-year search for an attorney, HBI was referred to the Public Interest Law Center of Philadelphia (PILCOP). In April 1985, PILCOP agreed to represent HBI and enlisted the assistance of the Tulsa, Oklahoma firm of Bullock & Bullock to serve as local counsel.2
Upon securing representation, HBI filed a class action suit on behalf of the residents of HMC. The complaint sought in-junctive and declaratory relief for eonstitu-tional and statutory violations by the state in its treatment of the residents.3 After nearly two-and-one-half years of litigation, the district court entered a judgment for Plaintiffs, expressly finding them to be the prevailing party and entitled to attorneys’ fees.
Plaintiffs’ attorneys submitted an application for attorneys’ fees which, in addition to requesting a fee based on the lodestar, requested an enhancement based on their superior performance, their acceptance of the case contingent upon receipt of fees awarded by the court, the risks involved, and the unpopularity of the case. The district court denied the request for an enhancement and awarded fees to Plaintiffs’ attorneys based solely on the lodestar. In denying the Plaintiffs’ request for a contingency enhancement, the district court reasoned:
the court is without evidence relating to the relevant market upon which it could adequately employ a “risk multiplier.” Further, the risk in this particular case, when viewed from the outset of the litigation, was not particularly great in light of existing federal law and the deplorable conditions at The Hissom Memorial Center. Some measure of success was fairly certain; the risk was the extent of the remedy the court would afford. To the extent that this risk lay in assembling the evidence and persuading the court of the extent of relief warranted, both these risks are fully addressed by the rate awarded the lawyers and the [1355]*1355number of hours this court deems reasonable in preparation of the case. Thus, the court can conclude only that a contingency enhancement is not appropriate in this case.
PILCOP subsequently settled the issue of its fees with Defendants. This appeal relates solely to Bullock’s fee application and specifically to the district court’s denial of a contingency enhancement.4
We review the district court’s award of attorney fees for an abuse of discretion. Smith v. Freeman, 921 F.2d 1120, 1122 (10th Cir.1990). Underlying factual findings will only be upset when clearly erroneous. Iqbal v. Golf Course Superintendents Ass’n, 900 F.2d 227, 228 (10th Cir.1990). However, a district court’s statutory interpretation or legal analysis which provides the basis for the fee award is reviewable de novo. Supre v. Ricketts, 792 F.2d 958, 961 (10th Cir.1986).
Congress has recognized that the amount of fees awarded under the Civil Rights Attorney’s Fees Awards Act of 1976 should be “adequate to attract competent counsel, but ... not produce windfalls to attorneys.” S.Rep. No. 1011, 94th Cong., 2d Sess., at 6 (1976), reprinted in 1976 U.S.C.C.A.N. 5908, 5913. See also H.R.Rep. No. 1558, 94th Cong., 2d Sess., at 9 (1976). In Hensley, the Supreme Court stated that the lodestar is “[t]he most useful starting point for determining the amount of a reasonable fee....” 461 U.S. at 433, 103 S.Ct. at 1939.5 More recently, the Supreme Court has considered the lodestar to be the presumptively reasonable fee:
A strong presumption that the lodestar figure—the product of reasonable hours times a reasonable rate—represents a “reasonable” fee is wholly consistent with the rationale behind the usual fee-shifting statute.... These statutes were not designed as a form of economic relief to improve the financial lot of attorneys, nor were they intended to replicate exactly the fee an attorney could earn through a private fee arrangement with his client. Instead, the aim of such statutes was to enable private parties to obtain legal help in seeking redress for injuries resulting from the actual or threatened violation of specific federal laws. Hence, if plaintiffs ... find it possible to engage a lawyer based on the statutory assurance that he will be paid a “reasonable fee,” the purpose behind the fee shifting statute has been satisfied.
Delaware Valley I, 478 U.S. at 565, 106 S.Ct. at 3098 (holding that quality of counsel's performance is not a basis to adjust the lodestar as this is normally reflected in the reasonable hourly rate). See also Blum, 465 U.S. at 897-900, 104 S.Ct. at 1548-49 (“novelty and complexity of the issues,” “special skill and experience of counsel,” “quality of representation,” and “results obtained” are presumably reflected and therefore generally subsumed within the lodestar and consequently will not justify an enhancement).
[1356]*1356Nonetheless, “[t]he product of reasonable hours times a reasonable rate does not end the inquiry [because] there remain other considerations that may lead the district court to adjust the fee upward or downward.” Hensley, 461 U.S. at 434, 103 S.Ct. at 1940. Accordingly, we have looked to the factors set forth in Johnson v. Georgia Highway Express, Inc., 488 F.2d 714, 717-19 (5th Cir.1974) (cited with approval in S.Rep. No. 1011, at 6, reprinted in 1976 U.S.C.C.A.N. at 5913; H.R.Rep. No. 1558, at 8), to guide the district court’s calculation and our review of a reasonable attorney’s fee under the statute. See Ramos v. Lamm, 713 F.2d 546, 552 (10th Cir.1983). See also Blanchard v. Bergeron, 489 U.S. 87, 93, 109 S.Ct. 939, 944, 103 L.Ed.2d 67 (1989) (“Johnson's ‘list of 12’ ... provides a useful catalog of the many factors to be considered in assessing the reasonableness of an award of attorney’s fees”); Hensley, 461 U.S. at 434 n. 9, 103 S.Ct. at 1940 n. 9 (“[t]he district court also may consider other factors identified in Johnson ... though it should note that many of these factors usually are subsumed within the initial calculation of hours reasonably expended at a reasonable rate”). Although one of these factors is “[w]hether the fee is fixed or contingent,” Johnson, 488 F.2d at 718, it is not entirely clear whether Congress intended contingency to be a factor in determining whether a fee is otherwise reasonable. See Pennsylvania v. Delaware Valley Citizens’ Council (Delaware Valley II), 483 U.S. 711, 723, 107 S.Ct. 3078, 3085-86, 97 L.Ed.2d 585 (1987) (plurality opinion) (“there is nothing in Johnson to show that this factor was meant to reflect the contingent nature of prevailing in the lawsuit as a whole”). See also Blanchard, 489 U.S. at 99, 109 S.Ct. at 947 (Scalia, J., concurring) (criticizing reliance on Johnson factors to determine congressional intent). Thus, while it is “doubtful that the legislative history [of § 1988] supports the use of this factor” in determining whether an enhancement to the lodestar is appropriate,” Delaware Valley II, 483 U.S. at 723, 107 S.Ct. at 3085 (plurality opinion), we and every other circuit that has considered the issue, with the exception of the D.C. Circuit, have recognized the availability of contingency enhancements in the calculation of reasonable fees under statutory fee shifting statutes such as § 1988. See Ramos, 713 F.2d at 558. See also Delaware Valley II, 483 U.S. at 717 n. 4, 107 S.Ct. at 3082 n. 4 (plurality opinion) (cases cited therein). But see King v. Palmer, 950 F.2d 771, 784 (D.C.Cir.1991) (en banc), petition for cert. filed, 60 U.S.L.W. 3617 (U.S. Feb. 21, 1992) (No. 91-1370).
While we have recognized the availability of contingency enhancements, we have “viewed [them] with caution.” Ramos, 713 F.2d at 558. “[W]hen determining whether to give a contingency bonus, the trial court should view the litigation as it reasonably should have appeared to the lawyers at the outset of the litigation.” Id. Accord Delaware Valley II, 483 U.S. at 729, 107 S.Ct. at 3088 (plurality opinion) (citing Ramos). In Ramos we held that because “the plaintiffs had little risk of not prevailing on some of the issues [and] [t]he real controversies and uncertainties lay in litigating the remedy ... some measure of success on the merits was fairly certain, and no bonus for contingency [was] warranted....” Id. Thus, we evaluate contingency enhancements based on the risk presented by the particular case at hand.6
Our practice of evaluating contingency enhancements based on the risks of the particular case finds support in Blum v. Stenson and the Delaware Valley II7 plu[1357]*1357rality opinion. In Blum, the Supreme Court held that an “upward adjustment for the contingent nature of the litigation was unjustified” in the particular case before it because the fee applicant failed to “identify any risks associated with the litigation or claim that the risk of nonpayment required an upward adjustment to provide a reasonable fee.”8 465 U.S. at 901, 104 S.Ct. at 1550. Similarly, in Delaware Valley II, to the extent that the plurality would permit a contingency enhancement,9 it “should be reserved for exceptional cases where the need and justification for such enhancement are readily apparent and are supported by evidence in the record and specific findings by the courts.” 483 U.S. at 728, 107 S.Ct. at 3088 (plurality opinion) (citing Blum, 465 U.S. at 898-901, 104 S.Ct. at 1548-50) (footnote omitted). In determining that the enhancement was not appropriate to the case before the Court, the Delaware Valley II plurality recognized that “[t]he risk of nonpayment should be determined at the beginning of the litigation,” id., 483 U.S. at 729, 107 S.Ct. at 3088 (plurality opinion) (citing Lewis v. Coughlin, 801 F.2d 570, 576 (2d Cir.1986); Ramos, 713 F.2d at 558), and that “ ‘[t]he test ... should be an objective one based on the likely response of the bar to the case’s pretrial merits, rather than on the judge’s subjective opinion of the merits.’ ” Id., 483 U.S. at 730 n. 11, 107 S.Ct. at 3089 n. 11 (plurality opinion) (quoting Lewis, 801 F.2d at 575). The Delaware Valley II plurality stated:
We have alluded to the uncertainties involved in determining the risk of not prevailing and the burdensome nature of fee litigation. We deem it desirable and an appropriate application of the statute to hold that if the trial court specifically finds that there was a real risk-of-not-prevailing issue in the case, an upward adjustment of the lodestar may be made, but, as a general rule, in an amount no more than one-third of the lodestar. Any additional adjustment would require the most exacting justification. This limitation will at once protect against windfalls for attorneys and act as some deterrence against bringing suits in which the attorney believes there is less than a 50-50 chance of prevailing. Riskier suits may be brought, and if won, a reasonable lodestar may be awarded, but risk enhancement will be limited to one-third of the lodestar, if awarded at all.
Id. (plurality opinion) (emphasis added).
In the present case, the district court, consistent with Ramos as well as our interpretation of the Delaware Valley II plurality opinion and Blum, relied in part on the risk of this particular case in declining to enhance the lodestar based on the contingency.10 Plaintiffs contend that [1358]*1358the district court should have applied the standard set forth in Justice O’Connor’s concurring opinion in Delaware Valley II. Under this standard, “compensation for contingency must be based on the difference in market treatment of contingent fee cases as a class, rather than on an assessment of the ‘riskiness’ of any particular case.”11 Id. at 731, 107 S.Ct. at 3089 (O’Connor, J., concurring). See also id. at 745-46, 107 S.Ct. at 3097 (Blackmun, J., dissenting). Justice O’Connor expressly rejected the award of any enhancement based on “legal” risks or risks peculiar to the case, reasoning that compensation for such risks should be reflected in the lodestar. Id. at 734, 107 S.Ct. at 3091 (O’Connor, J., concurring).
Several circuits considering the issue in the wake of Delaware Valley II have adopted Justice O’Connor’s opinion as setting forth the governing standard. See Morris v. American Nat’l Can Corp., 952 F.2d 200, 204 (8th Cir.1991); Islamic Ctr. v. City of Starkville, 876 F.2d 465, 471 (5th Cir.1989); Lattimore v. Oman Constr., 868 F.2d 437, 439 n. 4 (11th Cir.1989) (per curiam); Fadhl v. City & County of San Francisco, 859 F.2d 649, 650 n. 1 (9th Cir.1988) (per curiam); Blum v. Witco Chem. Corp., 829 F.2d 367, 379-82 (3d Cir.1987); Spell v. McDaniel, 824 F.2d 1380, 1404 (4th Cir.1987), cert. denied, 484 U.S. 1027, 108 S.Ct. 752, 98 L.Ed.2d 765 (1988). While we have not squarely addressed the issue, our post-Delaware Valley II decisions suggest that the Delaware Valley II plurality opinion controls. See Smith, 921 F.2d at 1123; Wulf v. City of Wichita, 883 F.2d 842, 875-76 (10th Cir.1989). Accord Skelton v. General Motors Corp., 860 F.2d 250, 258 (7th Cir.1988), cert. denied, 493 U.S. 810, 110 S.Ct. 53, 107 L.Ed.2d 22 (1989); Friends of the Earth v. Eastman Kodak Co., 834 F.2d 295, 298 (2d Cir.1987). As the issue is now squarely before us, we reaffirm the standard set forth in Ramos and approved of by the Delaware Valley II plurality, and we decline to adopt Justice O’Connor’s concurring opinion in Delaware Valley II as the law of this Circuit.
While we recognize that we are swimming up a stream of authority from our sister circuits, we are persuaded by the D.C. Circuit’s reasoning in King as to why Justice O'Connor’s opinion is not controlling.12 As King recognizes, the rationale for following Justice O’Connor’s concurring opinion in Delaware Valley II is derived from Marks v. United States, 430 U.S. 188, 97 S.Ct. 990, 51 L.Ed.2d 260 (1977), in which the Supreme Court held that “[w]hen a fragmented Court decides a case and no single rationale explaining the result enjoys the assent of five Justices, the holding of the Court may be viewed as that position taken by those Members who concurred in the judgments on the narrowest grounds.” Id. at 193, 97 S.Ct. at 993. Like the D.C. Circuit, we do not read Justice O’Connor’s opinion as concurring in the judgment on the narrowest grounds. See King, 950 F.2d at 793.
Specifically, Justice O’Connor would compensate for contingency based on the difference in market treatment of contingency cases as a class. Delaware Valley II, 483 U.S. at 731, 107 S.Ct. at 3089 (O’Connor, J., concurring). We have no doubt that contingency fee cases “typically generates [1359]*1359fees (if at all) substantially in excess of their more conservative counterparts,” such as hourly retainer cases, in order to compensate the attorney for the risk of losing and not being paid. In re Western Real Estate Fund, 922 F.2d 592, 597 (10th Cir.1990), modified, Abel v. West, 932 F.2d 898 (10th Cir.1991). See also Venegas v. Mitchell, 495 U.S. 82, 110 S.Ct. 1679, 109 L.Ed.2d 74 (1990) (enforcing attorney contingency fee agreement providing attorney with fees in excess of $400,000 even though statutory fee award under § 1988 amounted to only $75,000). Thus, if the market treatment of contingency cases as a class were the standard for determining whether a contingency enhancement is appropriate, we believe that a contingency enhancement would have to be given in every case. We believe that when Justice O’Connor spoke of compensating for the contingency based on the difference in market treatment of contingency cases as a class, she was speaking of the amount of compensation once it was determined that a contingency enhancement was proper. See King, 950 F.2d at 777 (whether enhancement is available at all is an “analytically distinct question[ ]” from the calculation of its amount).
For Justice O’Connor, the relevant inquiry in determining whether a contingency enhancement is warranted is whether the prevailing party “would have faced substantial difficulties in finding counsel in the local or other relevant market,” without the contingency enhancement. Delaware Valley II, 483 U.S. at 733, 107 S.Ct. at 3091 (O’Connor, J., concurring). On this point, she is in agreement with the plurality. See id. at 731, 107 S.Ct. at 3089 (plurality opinion). However, the plurality would also require, to the extent that it would permit contingency enhancements at all, a finding that there was “a real risk-of-not-prevailing issue in the case....” Id. at 730, 107 S.Ct. at 3089 (plurality opinion). Thus, we view the plurality opinion and Justice O’Connor’s concurring opinion as setting forth entirely different standards and, therefore, do not believe that Justice O’Connor’s position “posits a narrow test to which the plurality must necessarily agree as a logical consequence of its own broader position.” King, 950 F.2d at 782. Indeed, we read the plurality opinion as setting forth a narrower standard because it requires an initial showing of the risk peculiar to the particular case. As the D.C. Circuit recognized:
When ... one opinion supporting the judgment does not fit entirely within a broader circle drawn by the others, Marks is problematic. If applied in situations where the various opinions supporting the judgment are mutually exclusive, Marks will turn a single opinion that lacks majority support into national law. When eight of nine Justices do not subscribe to a given approach to a legal question, it surely cannot be proper to endow that approach with controlling force, no matter how persuasive it may be.
Id. at 782. See also Dague v. City of Burlington, 935 F.2d 1343, 1360 (2d Cir.1991) (recognizing “the anomaly of the views of one justice, with whom no one concurs, being the law of the land, where the Court is so divided on an issue and where there is no majority opinion at all”), cert. granted, — U.S. —, 112 S.Ct. 964, 117 L.Ed.2d 130 (1992). Accordingly, we do not believe that Justice O’Connor’s opinion overrules Ramos which, like the Delaware Valley II plurality, permits a contingency enhancement only when the particular case presents a real risk of not prevailing. See Dague, 935 F.2d at 1360 (“conclud[ing] that the issue of whether and when a contingency enhancement is warranted are open issues for the Supreme Court yet to decide” and reaffirming Second Circuit’s pre-Delaware Valley II approach).
While we are no less “loath to establish a rule that would inevitably result in attorney’s fee litigation in all contingency fee cases” than we have been in the past, Smith, 921 F.2d at 1123, our failure to articulate governing standards in light of the divergent views of the members of the Supreme Court in Delaware Valley II inevitably invites the very attorney’s fee litigation against which the Supreme Court has cautioned. See Hensley v. Eckerhart, 461 [1360]*1360U.S. 424, 437, 103 S.Ct. 1933, 1941, 76 L.Ed.2d 40 (1983) (“request for attorney’s fees should not result in a second major litigation”). Given the confusing state of the law, we are inclined to clarify the applicable standards in this Circuit to determine contingency enhancements. First, we reaffirm Smith, Wulf and Ramos by recognizing that a contingency enhancement may be given, notwithstanding that four members of the Supreme Court have stated that such enhancements are never appropriate. Delaware Valley II, 483 U.S. at 727, 107 S.Ct. at 3087 (plurality opinion). Indeed, five members of the Court have recognized contingency as an appropriate factor to consider in determining the reasonableness of a fee. Id. at 731, 107 S.Ct. at 3089 (O’Connor, J., concurring); id. at 735, 107 S.Ct. at 3091 (Blackmun, J., dissenting). While the D.C. Circuit has declined to follow the view of five members of the Court, holding that “contingency enhancements will not be available in this Circuit,” King, 950 F.2d at 784, we cannot follow its reasoning because to do so would require us to overrule precedent in this Circuit both before and after Delaware Valley II which recognizes that contingency enhancements may be given. See Smith, 921 F.2d at 1123; Ramos, 713 F.2d at 550. This would be clearly beyond our authority. United States v. Spedalieri, 910 F.2d 707, 710 n. 3 (10th Cir.1990) (three-judge panel cannot overrule circuit precedent).
We reaffirm our position that an enhancement for a contingency must be viewed with caution, Ramos, 713 F.2d at 558, and indeed, we believe that it is appropriate only in “exceptional cases.” Delaware Valley II, 483 U.S. at 728, 730, 107 S.Ct. at 3088, 3089 (plurality opinion). We believe that an “exceptional case” is one in which prior to the litigation, the attorney for the prevailing party was confronted with a “real risk of not prevailing.” Id. at 729-30, 107 S.Ct. at 3088-89 (plurality opinion). This inquiry is twofold. First, the district court should consider the law in effect at the time the suit was filed in light of the claims presented in the lawsuit. To the extent that the law was unsettled, a district court could properly conclude in the exercise of its discretion that the attorney faced a real risk of not prevailing. Second, the district court should consider the facts available to the attorney at the time the suit was filed. To the extent that the outcome of the suit was dependant on the resolution of a disputed material fact, the district court could properly conclude in the exercise of its discretion that the attorney faced a real risk of not prevailing. As to both the legal and factual inquiry, we reiterate that the burden of proving a real risk of not prevailing is on the party seeking the enhancement to the lodestar. Blum, 465 U.S. at 898, 104 S.Ct. at 1548.
Once a real risk of not prevailing has been demonstrated, the party seeking to enhance the lodestar must also come forward with evidence that, absent an enhancement, “the plaintiff would have faced substantial difficulties in finding counsel in the local or other relevant market.” Delaware Valley II, 483 U.S. at 731, 107 S.Ct. at 3089 (plurality opinion) (footnote omitted); id. at 733, 107 S.Ct. at 3091 (O’Connor, J., concurring). This may be shown by evidence that the plaintiff encountered actual difficulty in securing counsel, see id. at 731 n. 12, 107 S.Ct. at 3089 n. 12 (plurality opinion), as was shown in this case, or, as was also done in this case, by showing that lawyers in the relevant market, generally will not take cases on contingency absent some guarantee of an enhancement. See id. at 733-34, 107 S.Ct. at 3090-91 (O’Connor, J., concurring).
In the present case, despite Plaintiffs’ showing that they faced actual difficulty finding counsel, and that lawyers in the relevant market generally will not take cases on contingency absent an enhancement, the district court determined that Plaintiffs’ attorneys were not confronted with a real risk of not prevailing given the conditions at Hissom Memorial Center and the state of the law. Plaintiffs have not pointed to anything in the record which would lead us to conclude that this finding was clearly erroneous.
AFFIRMED.