WILL, Senior District Judge.
This is an appeal from the district courts’ awards of attorney’s fees in four social security cases. The claimants were granted awards after remand from the district courts to the Secretary of Health and Human Services (the “Secretary”). In each case, counsel had entered into a contingency agreement with the social security claimant. The Secretary opposed the fees as excessive, but the district courts awarded close to the full amounts requested pursuant to the contingency agreements. This appeal followed. We affirm two of the cases and remand two of them for additional findings.
I. Factual Background.
In each of the four cases, the claimant’s application for social security benefits was first denied by the Secretary and a complaint challenging that decision was filed in district court. The claimants, upon remand to the Secretary, were all ultimately successful in obtaining benefits. In the Gru-ber, Green and Marshall cases, the remands by the district courts were based in part on the revised criteria for mental impairments which were promulgated in the Social Security Disability Benefits Reform Act of 1984, 98 Stat. 1794, 1801-1802, Pub. L. No. 98-460 (“1984 Disability Amendments”).
The attorney in all four cases was Craig A. Fobes whose Wisconsin practice from 1981 to 1986 consisted mostly of social security cases. He has handled such cases since August of 1974. He has regularly received difficult social security cases by referral from other attorneys in Madison and Eau Claire. The McGuire case is one such case received upon referral.
A. McGuire v. Sullivan.
The trial court granted final judgment in June 1987 awarding Mr. McGuire Title II [976]*976benefits. In July of 1987, Mr. Fobes request for attorney’s fees for his work on the case under the Equal Access to Justice Act (“EAJA”), 28 U.S.C. § 2412 was denied, because the position of the Secretary in contesting the benefits was found by the court to be substantially justified. As a result, in September 1987 Mr. Fobes requested that the court award him the lesser of $14,458.72 or 25% of Mr. McGuire’s past-due benefits. In support of this request, Mr. Fobes referred to the contingency agreement made with Mr. McGuire and showed the court a record of 42.74 hours of legal service before the court. The Secretary opposed the fees as excessive, but on September 25, 1987, the court awarded $13,533.12.
The Secretary filed a motion to alter or amend the judgment which the court denied on November 16, 1987. In denying that motion and upholding the fee award, the court noted the factors laid out in Blankenship v. Schweiker, 676 F.2d 116 (4th Cir.1982) and also relied on the decision in Andrews v. Bowen, 640 F.Supp. 1001 (W.D.N.C.1986), vacated, 818 F.2d 28 (4th Cir.1987).
The district court distinguished from social security fee request cases the Supreme Court case of Pennsylvania v. Delaware Valley Citizens’ Council, 483 U.S. 711, 107 S.Ct. 3078, 97 L.Ed.2d 585 (1987) in which the plurality limited the degree to which, in certain cases, a fee may be multiplied above a reasonable hourly rate. The district court pointed out that the Delaware Valley principles were applied to a fee-shifting statute, which requires that the fees be paid by the losing party. In contrast, fees under § 206(b) of the Social Security Act are “awarded by the willing client who has entered into a contingency fee arrangement and thereafter has approved the arrangement.” McGuire v. Bowen, No. 85-C-268-S, slip op. at 5 (W.D. Wis. Sept. 24, 1987).
The district court listed as factors in support of the fee request, the “tremendous success achieved,” ibid., the skill of Mr. Fobes, the role of the contingency contract in securing legal representation for Mr. McGuire and the fact that the benefits would continue throughout Mr. McGuire’s lifetime.
B. Gruber v. Sullivan.
Mr. Fobes took Ms. Gruber’s case after another attorney, experienced in the Social Security area, had declined to take it because of its difficulty. Ms. Gruber’s case was remanded on November 15, 1984 for reconsideration of her mental impairments in light of the 1984 Disability Amendments. Mr. Fobes filed objections to the remand arguing in part that the question of whether an earlier application filed by Ms. Gru-ber had been reopened had to be resolved apart from whether she was disabled under the new amendments. Record at 14, pp. 2-3. Afterwards, Fobes and counsel for the Secretary stipulated to the reopening which made it possible for Ms. Gruber to receive three extra years of benefits. Mr. Fobes’ argument that the Secretary had violated Ms. Gruber’s rights pursuant to a court order in the class action, Mental Health Ass’n v. Schweiker, 554 F.Supp. 157 (D.Minn.1982), aff'd in part and modified in part sub nom., Mental Health Ass’n v. Heckler, 720 F.2d 965 (8th Cir.1983), was adopted by Magistrate Groh as a separate basis for the remand. Order of March 25, 1988, Record at 25.
In October 1986, Ms. Gruber was found disabled and in June the district court entered judgment awarding her Title II benefits. Mr. Fobes filed a request for attorney’s fees for 42.37 hours of court-related legal services on the case, in the amount of $9,486.77 or 25% of Title II and Title XYI benefits. The court allowed the fees over the objection of the Secretary on November 18, 1988, in a decision which is published at 673 F.Supp. 970 (W.D.Wis.1987). In that opinion, the court referred to Judge Sha-baz’ earlier fee decision in the McGuire case. In addition, the court applied the Blankenship factors to the case at hand. The district court found the case to be a difficult one due to “roadblocks” raised by the Secretary and the unusual disability of Ms. Gruber and noted the skill of her attorney, the successful result achieved and the reputation and ability of the attorney. The [977]*977district judge pointed out that the case had been taken by Mr. Fobes only after it had been declined by another attorney, that the fee was based on a contingency agreement and it was similar to other awards by district courts. In addition to the $9,486.77, the district court awarded $3,177.75 in attorney’s fees to Ms. Gruber pursuant to the EAJA. The EAJA award reduces to $6,309.02 the amount to be paid to Mr. Fobes out of Ms. Gruber’s past-due benefits.1
C. Green v. Sullivan.
Adrian Green’s mother applied for Title XVI disability benefits on his behalf, but the application was denied at every step of the appeal process by the Secretary. Again, Mr. Fobes accepted the case after another attorney had refused it, Suppl. App., p. 23C, and filed a claim in forma pauperis in federal district court on May 30,1985. Mr. Fobes persuaded counsel for the Secretary to stipulate to a remand pursuant to the 1984 Disability Amendments. The Secretary found Green disabled and the district court entered judgment in July 1987. In October 1987, Mr. Fobes requested $3,489.79 for attorney’s fees for 3.24 hours of work on the Green case. The court allowed the fee, following the principles announced in Gruber and stating that the time spent representing the claimant in federal court alone was not an adequate measure for fees in a social security case.
D. Marshall v. Sullivan.
After the Secretary denied Lyle Marshall’s application for Title XVI benefits, Mr. Fobes filed on his behalf a complaint in forma pauperis on September 18, 1984. The district court, sua sponte, remanded the case to the Secretary for reconsideration pursuant to the 1984 Disability Amendments. Mr. Marshall was found disabled and the district court entered judgment on March 26, 1987. Mr. Fobes requested in November 1987 that the court allow him $5,600 or 25% of the claimant’s past benefits for 5.6 hours of court-related services. The court approved the fee request and again announced that it was following the principles set out in Gruber.
II. Standard of Review.
In reviewing a fee award, we defer to “the district court’s superior understanding of the litigation and the desirability of avoiding appellate review of what are essentially factual matters.” Hensley v. Eckerhart, 461 U.S. 424, 437, 103 S.Ct. 1933, 1941, 76 L.Ed.2d 40 (1983). Factual matters are, of course, reviewed under a clearly erroneous standard. Fed.R.Civ.P. 52(a). We reverse a fee award decision, only upon finding an abuse of discretion. Hagge v. Bauer, 827 F.2d 101, 111-12 (7th Cir.1987). Therefore, “[i]f reasonable persons could differ over the trial court’s view, then the appellate court will not be able to find an abuse of discretion.” Munson v. Friske, 754 F.2d 683, 696 (7th Cir.1985).
III. Analysis.
A. Whether a fee based on a contingency agreement is reasonable.
Section 206(b) of the Social Security Act (“the Act”), codified at 42 U.S.C. § 406(b), governs the award of attorney’s fees under Title II of the Act and limits them to twenty-five percent of past-due benefits. The relevant part of the section provides:
(1) Whenever a court renders a judgment favorable to a claimant under this title who was represented before the court by an attorney, the court may determine and allow ... a reasonable fee [978]*978for such representation, not in excess of 25 percent of the total of the past-due benefits ... and the Secretary may, notwithstanding the provisions of section 205(i), certify the amount of such fee for payment to such attorney out of, and not in addition to, the amount of such past-due benefits.
42 U.S.C. § 406(b) (1982) (emphasis added). Attorney’s fee awards under Title XVI of the Act are not governed by any statutory provision, but courts have interpreted 42 U.S.C. § 405(i) (1982) (which directs the Secretary to certify payment of benefits), to authorize district judges to set attorney’s fees. See, e.g., Celebrezze v. Sparks, 342 F.2d 286, 288 (5th Cir.1965). See also Howard v. Bowen, 823 F.2d 185, 186 (7th Cir.1987) (finding that even though the district court did not have authority to order the Secretary to withhold benefits, the court was empowered to set them).2
The gist of the Secretary’s argument is that the starting place for a reasonable fee is the figure arrived at by multiplying the hours of legal work by a reasonable hourly wage (which yields the lodestar rate) and that the attorney’s fees awarded in the cases before us are not reasonable in light of the number of hours of work expended. In contrast, the district court in each case relied substantially on the contingent fee contract in effect between the clients and Mr. Fobes. The Secretary asserts that the district judges abused their discretion, because they ignored the lodestar calculation and thereby “rendered] the § 406(b) requirement that the court determine a reasonable fee a nullity and ignore[d] the fundamental purpose of the SSI program to provide subsistence level income to beneficiaries.” Brief of Defendant-Appellant at 19-20.
The Secretary relies upon several Supreme Court and Seventh Circuit cases decided in the context of fee-shifting statutes as support for his argument that the lodestar fee should only rarely be upwardly adjusted. He cites Blum v. Stenson, 465 U.S. 886, 898-99, 104 S.Ct. 1541, 1548-49, 79 L.Ed.2d 891 (1984), in which the Court held that the complexity, the novelty of legal issues and the special skills of counsel are not appropriate factors favoring an increase in the basic fee award. However, the Blum Court also wrote that “ ‘reasonable fees’ under § 1988 are to be calculated according to the prevailing market rates in the relevant community,” id. at 886, 104 S.Ct. at 1547, which must also include contingency rates.
The Supreme Court addressed the question of whether the lodestar rate should be enhanced because of the risk of nonpayment in the context of fee-shifting statutes in another case cited by the Secretary, Pennsylvania v. Delaware Valley Citizens’ Council, 483 U.S. 711, 107 S.Ct. 3078, 97 L.Ed.2d 585 (1987). The Secretary relies on the opinion of the four justice plurality for his argument against enhancement. Nevertheless, the majority holding of Delaware Valley is that the lodestar may be enhanced for contingency. Skelton v. General Motors Corp., 860 F.2d 250, 257 (7th Cir.1988). In Skelton, we wrote:
Justice O’Connor, casting the deciding vote, agreed with the plurality that the circumstances of Delaware Valley did not warrant the award of a risk multiplier. She agreed with the dissent, however, “that Congress did not intend to foreclose a consideration of contingency in setting a reasonable fee under fee-shifting provisions.” 107 S.Ct. 3089 [979]*979(O’Connor, J., concurring in part and concurring in judgment).
Ibid. Justice O’Connor argued in her concurrence that there had been no findings of fact in the case before the Court to support an enhancement. She wrote that such findings would include “the degree to which contingency is compensated in the relevant market.” Id. at 3089. She noted, in addition, that the district court had based the enhancement on the novelty of legal issues in the case, the type of risk which does not justify enhancement. To justify enhancement, she 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.
An even stronger case for enhancement can be made for social security fee awards than for fee awards pursuant to fee-shifting statutes. We have recognized that in a common fund case where the fee is taken from the plaintiffs’ award and, therefore, “there is no direct or immediate danger of unduly burdening the defendant, a court has more latitude in exercising its equitable powers to determine whether the plaintiff class should compensate its attorneys for their risk of nonpayment” than it would under the stringent Delaware Valley requirements. Skelton, 860 F.2d at 254. In the same way, in social security attorney’s fee requests such as the ones before us, the fees come out of the claimant’s benefits. We find support from the Third Circuit which distinguished fee-shifting cases from social security requests. Coup v. Heckler, 834 F.2d 313, 324 (3rd Cir.1987). Accord Rodriguez v. Bowen, 865 F.2d 739, 745-46 (6th Cir.1989) (en banc).
The contingency agreements entered into in the cases before us provide an additional reason in favor of enhancement. Unlike a fee-shifting case or a social security case where there is no fee agreement and where “the task of monitoring the attorney is shifted to the judge in separate litigation over fees if the plaintiff wins,” Delaware Valley, 107 S.Ct. at 3085, the contingency percentage taken from the successful claimant’s benefit award is negotiated privately between the attorney and the claimant who has an incentive at the time she seeks representation to find the best value in attorney services (the lowest percentage for good representation she can find given the difficulty of her case). The claimant and attorney place a value on the attorney’s risk of nonpayment. The percentage of the award agreed to by the parties should take account of the risk, as well as the ability of the attorney and estimated hours of legal work that the claim will require.
In the everyday contingency situation, the contract is normally enforced without modification, unless the clients are especially vulnerable to overreaching such as children or poorly equipped to protect their interests, as in class actions. See Dunn v. H.K. Porter Co., 602 F.2d 1105, 1111 (3rd Cir.1979); Rodriguez, 865 F.2d at 748 (Wellford, J., concurring). However, § 406 and the common law both require that the district court review the reasonableness of the award requested by the claimant’s attorney. Mr. Fobes does not dispute this.
A fee pursuant to a contingency contract is not per se reasonable. McKittrick v. Gardner, 378 F.2d 872, 875-76 (4th Cir.1967). Therefore, courts have reviewed social security fee requests according to a factor analysis which was originally taken from the fee-shifting context,3 such as the Blankenship v. Schweiker, 676 F.2d 116 (4th Cir.1982), analysis. The six factors must be weighed differently, however, when a contingency contract exists between the client and attorney. The Blankenship court noted that where a contingency contract exists, the court should not place primary emphasis on the hourly rate. [980]*980Blankenship, 676 F.2d at 118. In McKittrick, the court wrote:
[t]he contingency of compensation, whether it stems from an employment contract or results from the claimant’s indigency, is highly relevant in the appraisal of the reasonableness of any fee claim. The effective lawyer will not win all of his cases, and any determination of the reasonableness of his fees in those cases in which his client prevails must take account of the lawyer’s risk of receiving nothing for his services. Charges on the basis of a minimal hourly rate are surely inappropriate for a lawyer who has performed creditably when payment of any fee is so uncertain.
McKittrick, 378 F.2d at 875. See also Rodriguez, 865 F.2d at 746; Coup, 834 F.2d at 324; In re Horenstein, 810 F.2d 73, 75 (6th Cir.1986).
Where there was no contingency contract, we have found that the lodestar rate is the proper starting place, Lightfoot v. Walker, 826 F.2d 516, 520 (7th Cir.1987), but the rate may be enhanced by a risk factor. Skelton, 860 F.2d at 258. We have also held that “[t]here is no one correct formula for determining a fee award, and therefore the district court’s calculation is anything but an arithmetical exercise.” Tomazzoli v. Sheedy, 804 F.2d 93, 97 (7th Cir.1986). If courts regularly invalidated reasonable contingency agreements in favor of a lodestar fee, then attorneys would no longer enter into such agreements. Without the greater incentive for attorneys to take these cases stemming from the potential for an enhanced fee payment, claimants who have difficult cases and who cannot afford to guarantee payment might not be able to secure representation. Coup, 834 F.2d at 325; Dunn, 602 F.2d at 1111-12.
We recognize that other courts have determined that the lodestar rate is the proper starting place in analyzing social security fee requests and therefore have emphasized the reasonableness of the hourly rate of compensation from the award over the contingency agreed to by the parties. See, e.g., Starr v. Bowen, 831 F.2d 872, 874 (9th Cir.1987); Rohrich v. Bowen, 796 F.2d 1030, 1031-32 (8th Cir.1986); Bailey v. Heckler, 777 F.2d 1167, 1170 (6th Cir.1985), Nevertheless, we follow the McKittrick court’s reasoning that simply determining a reasonable hourly rate is inappropriate when an attorney is working pursuant to a reasonable contingency contract. A court may award a fee up to that provided in the contract so long as the court has reviewed its reasonableness. McKittrick, 378 F.2d at 875-76. Therefore, in reviewing social security fee applications where there is a contingency contract in effect, whether or not the proper starting place is a lodestar hourly rate, the contingency amount agreed to by the client should also be considered and a reasonable amount determined so long as it is no more than twenty-five percent.4
The Sixth Circuit en banc has recently addressed the question which is now before us and has decided that twenty-five percent of past-due benefits should be the starting place for all social security fee awards in order to reduce the court time spent reviewing fee requests. Rodriguez v. Bowen, 865 F.2d 739, 746 (6th Cir.1989) (“[o]ur aim is to provide a clearer and therefore more efficient framework for guidance ... in determining fee awards in social security cases”). In addition, when an attorney and client have entered into a contingency [981]*981agreement, “due deference should be given to this expression of the intentions of the parties” and therefore “if the agreement states that the attorney will be paid twenty-five percent of the benefits awarded, it should be given weight ordinarily accorded a rebuttable presumption.” Ibid. If the district court “chooses not to give effect to the terms of the agreement, it should state for the record the deductions being made and the reasons therefore.” Ibid. The court divided the potential deductions into two categories: “1) those occasioned by improper conduct of ineffectiveness of counsel; and 2) situations in which counsel would otherwise enjoy a windfall because of either an inordinately large benefit award or from minimal effort expended.” Ibid.
As an example of the first category of cases in which fees should be reduced, the Rodriguez court raised the issue of delay caused by the “procedural missteps” of an attorney which had resulted in an inflated award. Id. at 747. With regard to the second category, the court wrote:
[wjhere a case has been submitted on boilerplate pleadings, in which no issues of material fact are present and where no legal research is apparent, the benchmark twenty-five percent of awards fee would obviously be inappropriate. The reviewing courts should not hesitate to make reductions in such situations, and at the other end of the spectrum should only allow maximum fees for extensive effort on the part of counsel who have overcome legal and factual obstacles to the enhancement of the benefits awarded to his client.
Ibid.
Although we do not go as far as the Rodriguez court in finding a presumption in favor of a twenty-five percent contingency agreement, we agree with that court’s reasoning that we should follow a more efficient system for fee analysis and should defer to the parties’ intentions where reasonable. We agree that the fee agreement entered into by the parties should be the starting place for a court’s review but that amount may be reduced if appropriate. If the fee award amounts to a “windfall,” the court must carefully review the reasonableness of the contingency agreement. In all cases, the court should consider the reasonableness of the contingency percentage to make sure the attorney does not receive fees which are out of proportion to the services performed, the risk of loss and the other relevant considerations.
Although the courts below, in each case, did not blindly enforce the contingency contracts, it is not clear that they reviewed the extent of the contingency entered into in each case. The court must look for evidence of overreaching or failure to inform the client of her options. If, for instance, there is evidence that the client did not know she could negotiate other terms than a twenty-five percent contingency, then the contract should not be enforced. In addition, the court must review the strength of the claimant’s case in order to determine whether the percentage is reasonably related to the risk of loss before the Secretary.
We admitted, in Skelton, 860 F.2d at 258, the difficulty of objectively evaluating risk in light of the successful result already achieved, but found this factor insufficient to preclude risk enhancement in all cases. Since district judges review many of these cases, their experience should make it possible to recognize categories of easy and difficult cases. The deference to the contingency contract would be overcome, for example, by evidence that an attorney accepted an easy case pursuant to a full twenty-five percent contingency contract. The Blankenship factor approach should aid in determining the fairness of the contingency agreement, because absent a change in the law or the revelation of new evidence in favor of the claimant’s case the number of hours actually devoted to the case and the result obtained show whether or not a reasonable prediction was made.
In light of our understanding that contingency contracts are an important means for securing representation for social security claimants, we address the Secretary’s other arguments. He argues that a twenty-five percent contingency is unfair, when the attorney is fully compensated for [982]*982his time under the lodestar approach. Brief of Defendant-Appellant at 24. Since we find that a contingency award also compensates an attorney for taking the risk that he will not be paid, this argument has no weight. The Secretary also argues that contingency arrangements “penalize[] the Social Security [Title II] and SSI [Title XVI] claimants with the strongest cases, because they effectively subsidize other unsuccessful claimants who may not pay attorney fees at all.” Id. at 25. However, assuming that twenty-five percent contingency agreements are used as they should be, only for difficult cases, claimants with stronger cases should receive better terms. In addition, successful claimants who pay reasonable fees do not subsidize unsuccessful ones, but pay for the risk of loss in their own cases.
The Secretary cites in his favor several cases in which courts have refused to make awards based on contingency agreements, because “the amount of disability benefits to which a claimant is entitled on winning a case does not necessarily depend upon and frequently is not related to the skill or efforts of an attorney.” Brief of Defendant-Appellant at 16-17 (citing Bailey v. Heckler, 777 F.2d 1167, 1170 n. 3 (6th Cir.1985); Webb v. Richardson, 472 F.2d 529, 537 (6th Cir.1972); McKittrick v. Gardner, 378 F.2d 872, 874 (4th Cir.1967)). See also Rodriguez, 865 F.2d at 743 (distinguishing social security from personal injury cases).
The Secretary argues that the amount of benefits are determined by characteristics of the plaintiff that the attorney cannot change such as the amount of past wages. See 20 C.F.R. §§ 404.201-468, 416.401-435 (1988). However, although the rate at which the benefits are paid is determined by factors beyond the attorney’s control, the right to entitlement and the length of entitlement determine the total amount of the award. Several factors go into the determination of this length, including the date of onset of the disability, the date of application and the ending date of the disability. See 20 C.F.R. §§ 404.315, 404.316 (1988). Each of these factors must be proven. As a result, the length of disability period found by the Secretary will depend in part on the skill of a claimant’s attorney. A good attorney may, for example, be able to get an earlier unsuccessful application for benefits reopened, as was done in the Gruber case, so that the claimant will be entitled to more accrued benefits. Although we agree with Rodriguez that there is less potential for an attorney to enhance a social security award than a personal injury award (with punitive damages, pain and suffering, etc.), we do not agree that the social security award is not increased in many cases due to the skill of the attorney.
The Secretary also argues that the contingency fee arrangement is an abuse of discretion, because it rewards dilatory attorneys. He noted that “a rule was established [in the Sixth Circuit] to address this problem of windfalls to attorneys as a result of delay, regardless of fault. Dearing v. Secretary of Health and Human Services, 815 F.2d 1082, 1083 (6th Cir.1987).” Brief of Defendant-Appellant at 17. In Dearing, the court vacated a twenty-five percent contingency judgment because much of the award was due to the district court’s four-year delay in deciding the plaintiff’s motion for summary judgment. The reviewing court found that the twenty-five percent award should be based on the amount the client would have been awarded three months after the case was ready for submission. There is no evidence of such delay in any of the four cases we are reviewing.
The Secretary also argues that these fee awards should not be upheld, because “clients may not realize they have a choice in attorney fee matters or may not be aware how little time may actually be devoted in a given case.” Brief of Defendant-Appellant at 19. In support of this argument, the Secretary cites the finding of the court in McKittrick, 378 F.2d at 874, that claimants usually give their lawyers credit for the success in winning an award of benefits. As we have noted, we do not propose a blind approval of contingency fees. A district judge must review the risks in particular cases. In addition, we have no reason to believe, based on the [983]*983record, that claimants are always uninformed as to their choices in fee arrangements.
The Secretary makes a final argument against the need for contingency based on the alleged availability of attorneys. He cites the existence of “legal clinics, legal aid societies, and [attorneys] from the private sector” who will take on these cases “at far more modest fees than the court authorized here.” Brief of Defendant-Appellant at 29. However, the Supreme Court in Blum v. Stenson, 465 U.S. 886, 104 S.Ct. 1541, 79 L.Ed.2d 891 (1984), held that legal aid offices are entitled to market rate fees. The Secretary cites an Office of the Inspector General study which found that ninety-six percent of the interviewed claimants had no trouble hiring an attorney. However, neither the large number of attorneys nor the fact that claimants now do not have difficulty getting attorneys provides any evidence that claimants would be able to get attorneys if we were regularly to invalidate contingency agreements. Moreover, in the cases here involved, there is evidence that some of the claimants, at least, did have difficulty in securing counsel.
B. Whether the district courts abused their discretion in the cases before us.
The Secretary’s final line of argument is that even if some form of factor analysis were appropriate rather than focusing on the lodestar, the district courts erred in their analysis of the fee requests in these cases. He asserts that “[t]he district court must do more than merely recite a factor and conclusively decide that it applies to support an extraordinary multiplication of fees.” Brief of Defendant-Appellant at 27. The Blankenship factors include “time and labor required; skill required; contingency of fee; amount involved and result attained; experience, reputation, and ability of attorney; and awards in similar cases.” Blankenship, 676 F.2d at 118.
In order to show abuse, the Secretary repeats many of the arguments already made in favor of the lodestar approach. In addition, he argues that the court in McGuire erred by relying on Mr. Fobes’ notable success, since the denial of EAJA fees showed “that the case was a close one and substantial evidence was available to support a denial or an award of benefits.” Brief of Defendant-Appellant at 28. However, it is obvious that a close case is a risky one and a difficult one, since it could have gone either way.
The Secretary attempts to rebut the district courts’ reliance on the fact that the Gruber and Green cases were turned down before Mr. Fobes finally took them on by stating that “the reasons an attorney might decline a case are myriad.” Brief of Defendant-Appellant at 32. However, the record includes affidavits from the attorneys who refused the cases stating that they did not take the cases because of their difficulty. In short, none of the Secretary’s arguments establish that the trial courts abused their discretion. We do, however, have an additional concern as to whether the district courts sufficiently reviewed the reasonableness of the contingency agreements in each case.
C. Whether the contingency contracts were reasonable.
Although the contingency agreement should be given significant weight in fixing a fee, a district judge must independently assess the reasonableness of its terms. In the four cases before us, the district judges emphasized the contingency factor. However, in two of the cases the court did not give sufficient reasons why the Blankenship factors and the risk of loss in the case before it justified the contingency fee awarded.
1. McGuire v. Sullivan.
The district court, in McGuire, awarded somewhat less than twenty-five percent of past-due benefits. In support of the award, the court took into consideration the skill, experience and reputation of Mr. Fobes, the difficulties of the case, as well as the successful result obtained. In addition, the court focused on the need for contingency arrangements to guarantee representation for social security claim[984]*984ants. The court pointed out that there had been similar awards in cases “with similar results where lawyers have exhibited the skill, expertise and experience as that demonstrated by Mr. Fobes,” No. 85-C-268-S, slip op. at 6 (W.D.Wis. Nov. 16, 1987).
The district court’s analysis in its order denying Mr. Fobes’ motion for EAJA attorney's fees shows the risk and the difficulty of the case. The court wrote that although Mr. McGuire was successful on his claim, “there was also substantial evidence in the record to justify a contrary conclusion.” No. 86-C-268-S, slip op. at 6 (W.D.Wis. Oct. 9, 1987). In addition, the court found “that the Secretary had a solid and correct position on the facts,” which showed that his position was substantially justified. Ibid. Such findings show the high risk of loss in the case, even if they were not made with regard to Mr. Fobes’ contingency fee request. They are sufficient to show that the award was not an abuse of discretion.
2.Gruber v. Sullivan.
In Gruber, the court included in its factor analysis several conclusions as to the riskiness of the case. The court noted that the “case was a difficult one,” that “plaintiff’s disability is an unusual one,” the undesirability of the case shown by the fact that another attorney refused it and the skill of Mr. Fobes. 673 F.Supp. 970, 972 (W.D.Wis.1987). Although the 1984 Disability Amendments may have improved the chances of success of Ms. Gruber’s claim, the court found that it was a difficult one. The fact that another attorney had refused to take the case shows that it was risky. Mr. Fobes filed objections to the court’s sua sponte remand because of the delay caused by the arguably unnecessary remand and the fact that the remand for reconsideration of Ms. Gruber’s second application did not resolve the issue of benefits from the time her first application was filed. Apparently, the 1984 Disability Amendments did not make the Gruber case easy. The court, therefore, did not abuse its discretion by approving the twenty-five percent fee.
3. Green v. Sullivan.
In Green, the court outlined “plaintiff’s counsel’s skill, his experience, reputation, and ability, awards in similar cases, the contingency of the fee, and the amount involved in the result obtained,” No. 85-C-415-C, slip op. at 2 (W.D.Wis. Nov. 20, 1987), as factors in favor of the full twenty-five percent award. The court did not, however, give reasons why those factors justified such a large award in this case, such as whether or not this case was especially risky.
The fact that Mr. Fobes was awarded $3,489.79 for 3.24 hours of court-related work does not by itself show that the award is unreasonable. However, such a large award does require that the district judge make a careful review of the contingency agreement to determine whether or not the attorney has been granted an unreasonable windfall. See Rodriguez, 865 F.2d at 747. The small number of hours required to bring this case to final disposition is due in large part to the change in law caused by the passage of the 1984 Disability Amendments. The district court could, therefore, conclude that, although twenty-five percent was a reasonable percentage at the time the agreement was entered into, it is not reasonable in light of the change of law. The fact that another attorney had refused the case because of its difficulty is one factor which shows the risk of nonpayment at the time the case was taken on.
4. Marshall v. Sullivan.
The district court in Marshall based its approval of a full twenty-five percent fee on the skill of Mr. Fobes, “the uncertainty of victory in these cases,” No. 84-C-740-C, slip op. at 2 (W.D.Wis. Dec. 11, 1987), the fact that the claimant made no objection to the fee payment, the existence of a contingency agreement and the results obtained. However, as in Green there were no reasons given justifying the substantial award in this case as compared to other social security cases. As in Green, the fact that Mr. Fobes received $5,600 for 5.6 hours of court-related service shows that the district [985]*985court must carefully review the reasonableness of the twenty-five percent agreement in light of the risk of nonpayment.
IV. Conclusion
Because the district court in Gruber made specific findings regarding the difficulty, riskiness and other relevant factors which demonstrate that a twenty-five percent contingency contract is reasonable, we affirm the decision to award fees in that case. In McGuire, there are also sufficient findings in the court’s order denying EAJA fees and in that awarding fees to sustain the award as reasonable. However, in Green and Marshall, the district court opinions do not provide any analysis of the relevant factors in those cases, as compared to other contingency contract cases, which would justify a full twenty-five percent award of fees.
We must remand a case, when the district court opinion regarding fees does not explain the reasons for its conclusions. Lenard v. Argento, 808 F.2d 1242, 1247 (7th Cir.1987). See also Glass v. Secretary of Health and Human Services, 822 F.2d 19, 21 (6th Cir.1987) (district court must articulate reasons for § 406 awards to make meaningful review possible). Here, the courts provided reasons in support of an enhanced fee, but did not consider in Green and Marshall whether the risk in those cases merited a full twenty-five percent award. The skill and reputation of Mr. Fobes cannot justify that percentage in all cases, unless he only accepts difficult ones. A finding of riskiness is an essential one in granting a full twenty-five percent contingent fee award in a social security case. The trial court must determine whether the contingency percentage requested was reasonable in light of the risk, the hours of legal work necessary to complete the case and the skill and reputation of the attorney. We remand in Green (88-1143) and Marshall (88-1257) for that purpose.
Affirmed in part and remanded with instructions.