LEVIN H. CAMPBELL, Chief Judge.
The Secretary of Health and Human Services (“HHS”) appeals from the award of attorneys’ fees to plaintiffs in a suit that had challenged certain Social Security regulations. The fees were awarded by the district court under the Equal Access to Justice Act (“EAJA”). Enacted in 1980, EAJA provides that in some circumstances the United States shall be ordered by the court to pay the attorneys’ fees of a party that prevails against it. The relevant portion of EAJA provides,
Except as otherwise specifically provided by statute, a court shall award to a prevailing party other than the United States fees and other expenses ... incurred by that party in any civil action (other than cases sounding in tort), including proceedings for judicial review of agency action, brought by or against the United States in any court having jurisdiction of that action, unless the court finds that the position of the United States was substantially justified or that special circumstances make an award unjust.
28 U.S.C. § 2412(d)(1)(A) (Supp.1987).
Under EAJA, then, absent special circumstances, the government must foot the legal bills of its adversaries in civil cases (other than tort actions), but only if the adversaries “prevail” and if the govern-[1470]*1470merit’s position is not “substantially justified.”
I. BACKGROUND
The class action suit against HHS that resulted in the EAJA award contested HHS procedures for determining eligibility for Social Security disability benefits. Because the course of the underlying litigation bears directly on the attorneys’ fees issue, we must review it in some detail.
Two Social Security programs, Old Age Survivors and Disability Insurance (“OAS-DI”) and Supplemental Security Income (“SSI”), both provide for the payments of benefits to disabled persons.1 Under the Social Security Act (“the Act”), “disability” refers to the inability to do gainful work. A person is considered to be disabled only if “his physical or mental impairment or impairments are of such severity that he is not only unable to do his previous work but cannot, considering his age, education, and work experience, engage in any other kind of substantial gainful work which exists in the national economy_” 42 U.S.C. § 423(d)(2)(A) (1982). The plaintiffs brought suit in 1984, in order to challenge two aspects of HHS’s method of determining such disability, the “severity regulation” and the pre-1984 “combination of impairments regulation.”
1. The Severity Regulation.
The more important of the two challenges was directed at the HHS policy of screening out claimants with “non-severe” impairments at Step 2 of a five-step determination process that was introduced in 1978. The HHS regulations describe this screening step as follows:
You must have a severe impairment. If you do not have any impairment or combination of impairments which significantly limits your physical or mental ability to do basic work activities, we will find that you do not have a severe impairment and are, therefore, not disabled. We will not consider your age, education, and work experience_
20 C.F.R. § 404.1520(c) (1988) (emphasis supplied in last sentence). If a claimant is found at this step not to have a “severe impairment,” then his or her claim will not proceed to the remaining stages of the determination process, which include explicit consideration of age, education, and work experience. Plaintiffs contended that this regulation was inconsistent with the Act, because it allowed HHS to deny benefits to claimants without specifically considering their age, education, and work experience, as mandated by the Act. The district court accepted this argument. On motion for summary judgment, the court held that the Step 2 severity regulation was “inconsistent with the Social Security Act” and thus “invalid as written.” McDonald v. Heckler, 624 F.Supp. 375, 380 (D.Mass.1985). It accordingly enjoined the Secretary of HHS from enforcing the Step 2 regulation. Because the district court found the regulation to be invalid on its face, it did not specifically find that it was invalid “as applied.”
This court reviewed the district court’s judgment in McDonald v. Secretary of Health and Human Services, 795 F.2d 1118 (1st Cir.1986), which we will refer to as McDonald I. We found that the Step 2 severity regulation was valid if used by HHS as a de minimis screening step. Such a de minimis interpretation of the regulation had been provided by Social Security Ruling 85-28, which was issued in October 1985 “in response to the considerable judicial criticism of the severity regulation.” McDonald I at 1124. Although SSR 85-28 had been issued about a month before the district court’s order, the district court did not refer to it in its opinion. SSR 85-28 provided that a finding of “non-severe” is to be made at Step 2 only if “medical evidence establishes only a slight abnormality or combination of slight abnormalities which would have no more than a minimal effect on an individual’s ability to [1471]*1471work even if the individual’s age, education, or work experience were specifically considered_” In light of SSR 85-28, we vacated the district court’s injunction barring the Secretary from applying the severity regulation. We remanded the cases of the individual plaintiffs to the district court, id. at 1126, and we directed that the other class members2 exhaust their administrative remedies before seeking review in the district court. Id. Although we upheld the severity regulation as written, we expressed doubt about whether it had been properly applied before the issuance of SSR 85-28. Id. at 1124. A year after our opinion in McDonald I, the Supreme Court upheld the severity regulation as written. Bowen v. Yuckert, 482 U.S. 137, 107 S.Ct. 2287, 96 L.Ed.2d 119 (1987).
2. The “Combination of Impairments” Regulation.
Since 1967, the Social Security Act has required the Secretary to find a claimant disabled “if his physical or mental impairment or impairments are of such severity that he is not only unable to do his previous work, but cannot ... engage in any other kind of substantial gainful work.” 42 U.S.C. § 423(d)(2)(A) (emphasis supplied). In 1980, however, a new HHS regulation provided that HHS would “consider the combined effects of unrelated impairments only if all are severe[.]” 20 C.F.R. § 404.1522 (1981) (emphasis supplied). A 1984 amendment to the Social Security Act countermanded this interpretation, requiring the Secretary to consider “the combined effect of all of the individual’s impairments,” whether or not each would be deemed “severe” if considered separately. 42 U.S.C. § 423(d)(2)(C) (Supp.1987). The present plaintiffs challenged HHS’s refusal prior to December 1, 1984 (the amendment’s effective date) to consider the combined effect of impairments, claiming that this policy was in violation of the Act even before the amendment. The district court granted plaintiffs’ motion for summary judgment on this claim. The court directed HHS to consider on remand the combined effect of non-severe impairments when evaluating the claims of class members who had received a final decision from HHS before December 1,1984. McDonald v. Heckler, 629 F.Supp. 1138, 1140 (D.Mass.1986). On appeal, in McDonald I, we affirmed this order, noting that the Secretary’s rationale — that if one non-severe impairment does not significantly limit the ability to work, neither will the combination of two or more non-severe impairments— defied common sense. McDonald I, 795 F.2d at 1127.
The next stage of the litigation involved the proceedings on remand and their seque-lae. In considering these events it is useful to treat separately the two sets of plaintiffs in the case: the four individually named plaintiffs, and the thousands of unnamed members of the plaintiff class certified by the district court.
A. Individually Named Plaintiffs.
The four individually named plaintiffs were claimants for disability benefits whose claims had been denied as “non-severe” at Step 2 of the determination process. In McDonald I we remanded their cases to the district court for determination of whether “the proper standards, as set forth in Ruling 85-28 and this opinion” had been applied by the Secretary when their claims were denied. 795 F.2d at 1126. The district court, in turn, remanded the named plaintiffs’ claims to HHS, which in all four cases reversed its earlier rulings and awarded disability benefits. These determinations were not challenged on appeal.
B. Class Plaintiffs.
In June, 1985, in its first published opinion in this case, the district court had certified the following class:
All persons residing in Massachusetts who have filed or will file applications for disability benefits under Title II or Title XVI of the Social Security Act, and whose benefits have been or will be denied on the grounds that they do not have a severe impairment, pursuant to the policies set forth in [the Step 2 sever[1472]*1472ity regulation, the “combination of impairments” regulation, and associated Social Security Rulings].
McDonald v. Heckler, 612 F.Supp. 293, 299 (D.Mass.1985). The district court included in the class individuals who had filed claims but had not yet received a final decision from the Secretary. Citing Mathews v. Eldridge, 424 U.S. 319, 328-30, 96 S.Ct. 893, 899-900, 47 L.Ed.2d 18 (1976), the court ruled that the usual jurisdictional requirement of exhaustion of administrative remedies, see 42 U.S.C. § 406(g) (1982), was waived because these individuals’ complaints about HHS procedures were “wholly collateral to their individual claims for benefits.” 612 F.Supp. at 298. In McDonald I we did not discuss the propriety of the class certification. On remand, the district court ordered the Secretary, over the Secretary's jurisdictional objections, to notify the affected class members of their right to revive their claims and resume the administrative review procedure. McDonald v. Bowen, No. 84-2190-G, 1987 WL 42446 (D.Mass. Jan. 29, 1987). In a subsequent order, the district court established deadlines and detailed procedures for administrative review of the various claims. McDonald v. Bowen, No. 84-2190-G (D.Mass. Apr. 10, 1987).
The Secretary returned to this court with his jurisdictional objections to these orders. He argued that the district court never had jurisdiction over the claims of the class members because these plaintiffs had not obtained a “final decision of the Secretary,” as required under 42 U.S.C. § 405(g). In McDonald v. Secretary of Health and Human Services, 834 F.2d 1085 (1st Cir.1987) (to be referred to as McDonald II), we wrote that the jurisdictional question was “a close and perplexing one,” id. at 1089. Nevertheless, we affirmed both the district court’s jurisdiction and its judgment. We noted that we were now at the end — not the beginning — of the class action, that we had not suggested in our earlier opinion that the class was improperly certified, and that the appealed orders were “extremely narrow, being limited solely to affording an opportunity for equitable redress to class members who, in possible reliance upon the ongoing class litigation, may understandably have failed to perceive the need to pursue their individual claims.” Id. at 1091.
After our opinion in McDonald I, plaintiffs filed in the district court and in this court petitions for attorneys’ fees pursuant to EAJA for work done at the trial court and appellate levels. We remanded the fee petition for appellate work to the district court. The district court then granted plaintiffs’ motion to hold both fee petitions in abeyance pending further proceedings. After our opinion in McDonald II, plaintiffs filed in this court a supplemental fee petition for work done in connection with HHS’s second appeal. On March 24, 1988, we issued a one-sentence order denying this petition. Plaintiffs then filed a supplemental fee petition in the district court for their lawyers’ additional work at the district court level. At the request of the district court, plaintiffs filed an itemized fee petition showing the time their lawyers spent on each issue in the litigation.
II. THE DISTRICT COURT’S AWARD OF ATTORNEYS’ FEES UNDER THE EQUAL ACCESS TO JUSTICE ACT
As noted on page 1469, supra, EAJA provides that the government shall pay the attorneys’ fees of parties that prevail against it in civil litigation, if the court finds that the government’s position was not “substantially justified.” In considering plaintiffs’ applications for attorneys’ fees, the district court first found that they were indeed the “prevailing parties” for purposes of EAJA. Relying on the standard set forth in Nadeau v. Helgemoe, 581 F.2d 275, 278-79 (1st Cir.1978), the district court found that the plaintiffs had prevailed because they “clearly succeeded in obtaining a substantial portion of the benefit sought.” McDonald v. Bowen, 693 F.Supp. 1298, 1301 (D.Mass.1988).
The district court then turned to the question of whether the government’s position was “substantially justified.”3 Citing [1473]*1473both our opinion in McDonald I and the Supreme Court’s subsequent opinion in Bowen v. Yuckert, 482 U.S. 137, 107 S.Ct. 2287, 96 L.Ed.2d 119 (1987), the court found that HHS’s position was substantially justified with respect to the facial validity of the Step 2 severity regulation. But the court found that “[f]or all intents and purposes, the plaintiffs had prevailed on their challenge to the severity regulations as applied.” 693 F.Supp. at 1304 (emphasis added). The court found that the government had failed to show that it had been justified in its application of the regulation to claimants — presumably including the present plaintiffs — before the issuance of SSR 85-28. The court also found that on the pre-1984 “combination of impairments” issue, the government “has clearly failed to sustain its burden of proving that its position was reasonable.” 693 F.Supp. at 1302.
Because the district court found that plaintiffs had prevailed and that at least some of the government’s positions had not been justified, it ruled that plaintiffs were entitled to recover attorneys’ fees. The court then turned to the scope of these fees. Adopting the approach outlined by the Supreme Court in Hensley v. Eckerhart, 461 U.S. 424, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983) (applying the Civil Rights Attorney’s Fees Awards Act, 42 U.S.C. § 1988 (1982)), the district court focused on “the significance of the overall relief obtained by the plaintiff in relation to the hours reasonably expended on the litigation.” Hensley, 461 U.S. at 435, 103 S.Ct. at 1940 (quoted by the district court at 693 F.Supp. at 1305). Most of plaintiffs’ attorneys’ work was directed at the severity regulation, both as written and as applied. The work on these two fronts overlapped. By succeeding in their challenge to the regulation as applied, “the plaintiffs essentially succeeded in winning the relief they had sought.” Id. Accordingly, the district court awarded to plaintiffs fees for all of their attorneys’ billable hours for their work on the challenge to the severity regulations, as well as for their work on the “combination of impairments” issue.4 Id. Finally, the district court awarded fees for the work done by plaintiffs’ attorneys on the EAJA petition itself. Id. at 1305-06 n. 4. A total of $69,299.17 was awarded.5
III. STANDARD OF REVIEW
In Pierce v. Underwood, — U.S. —, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988), the Supreme Court held that a district court’s determination of whether the government’s position was “substantially justified” under EAJA can be reviewed only for abuse of discretion. In arriving at this deferential standard, the Court noted that as a matter of sound judicial administration the district court is typically better equipped than an appellate court to decide whether a government position was “substantially justified.” For example, the district court judge may “[by] reason of settlement conferences and other pretrial activities ... have insights not conveyed by the record.” Id. 108 S.Ct. at 2547. The Court observed that “[i]t is especially common for issues involving what can broadly be labeled ‘supervision of litigation,’ which is the sort of issue presented here, to be given abuse-of-discretion review.” Id. at 2546 n. 1. Finally, the Court asserted that the deferential review of a district court’s decision “will implement our view that ‘a request for attorneys’ fees should not result in a second [1474]*1474major litigation.’ ” Id. at 2549 (quoting Hensley v. Eckerhart, 461 U.S. at 437, 103 S.Ct. at 1941).
The Underwood decision dealt only with the appellate review of a district court’s determination of “substantial justification” under EAJA. But the logic of Underwood dictates that we apply the same deferential standard of review to the district court’s identification of a “prevailing party” and to its determination of the scope of an EAJA fee award. Indeed, the district court is probably in a better position to decide whether a party has “prevailed” (a decision that is based on an appreciation of the party’s litigation goals, strategies, and results) than it is to decide whether the government’s position was “substantially justified” (a decision that is more likely to turn on purely legal determinations). In addition, a district court will usually be better equipped than an appellate court to decide which of a party’s litigation efforts were reasonably related to the party’s success. Moreover, the goal of preventing attorneys’ fee petitions from becoming “second major litigations” suggests that we apply a deferential standard of review to all aspects of the district court’s EAJA determination. Thus, we review the district court’s award of attorneys’ fees under an abuse of discretion standard.
IV. “PREVAILING PARTY”
In deciding whether plaintiffs were “prevailing parties” in this litigation, the district court applied the standard that we set forth in Nadeau v. Helgemoe, 581 F.2d at 278-79: whether plaintiffs “succeeded] on any significant issue in litigation which achieve[d] some of the benefit the parties sought in bringing suit.” In Texas State Teachers Association v. Garland Independent School District, — U.S. —, —, 109 S.Ct. 1486, 1493, 103 L.Ed.2d 866 (1989), the Supreme Court endorsed the Nadeau standard as a test for deciding whether a civil rights plaintiff is a prevailing party under the Civil Rights Attorney’s Fees Awards Act, 42 U.S.C. § 1988. This standard is applicable to EAJA as well. When it enacted EAJA, Congress intended that its reference to “prevailing party” be interpreted in accord with the law that had developed under existing statutes, such as section 1988. See H.R.Rep. No. 96-1418, 96th Cong., 2d Sess. 11, reprinted in 1980 U.S.Code Cong. & Admin.News 4953, 4984, 4990 (cited by the district court at 693 F.Supp. at 1300-01). And in defining “prevailing party” under section 1988, the Supreme Court implied that its definition might apply to EAJA cases as well. Texas State Teachers Association, — U.S. at —, 109 S.Ct. at 1488 (“because of the importance of the definition of the term ‘prevailing party’ to the application of § 1988 and other federal fee shifting statutes, we granted certiorari”) (emphasis added).
The district court found that plaintiffs were prevailing parties because they “clearly succeeded in obtaining a substantial portion of the benefit sought.” McDonald, 693 F.Supp. at 1301. As a result of the litigation, the four named plaintiffs had their disability claims remanded to HHS, which reversed its earlier rulings and awarded them benefits. And “[mjore importantly,” the district court found, members of the plaintiff class—including some whose failure to exhaust their administrative remedies might have prevented them from obtaining any review—won the right to have HHS reconsider their disability status using the de minimis severity test as defined in SSR 85-28. Id. We add that the plaintiffs also prevailed in their challenge to the pre-1984 “combination of impairments” regulation.
The Secretary argues on appeal that plaintiffs were not “prevailing parties.” First, he reasons that when this court denied plaintiffs’ fee petition for their attorneys’ work on their second appeal, we decided in effect that they were not prevailing parties. But our summary denial contained no express determination to that effect nor is such a determination necessarily implied.
Second, the Secretary contends that this litigation did not provide to the four individual plaintiffs the relief they sought. All [1475]*1475they achieved through the litigation, the Secretary maintains, was a remand to HHS. Their success in obtaining benefits thereafter, the argument continues, may be attributable to the claimants’ worsening health or to changed administrative standards that were not related to this litigation. We are not persuaded. By means of the court-ordered remand, the individual plaintiffs received a second chance, not otherwise available, to obtain the benefits that had previously been denied them — and when given the opportunity, they actually obtained the benefits. Unlike Truax v. Bowen, 842 F.2d 995 (8th Cir.1988), a case relied on by the Secretary, the remand resulted not from a change in the law but from the courts’ concern that HHS had applied the severity test improperly. The district court did not abuse its discretion in finding that these plaintiffs were “prevailing parties.”6
Third, the Secretary argues that plaintiffs could not have “prevailed” because they lost on the claim that they argued most strenuously — that the Step 2 severity regulation was invalid on its face. Moreover, the Secretary continues, plaintiffs could not have prevailed on their “as applied” claim because neither the district court nor this court actually determined that the severity regulation was invalid as applied. In order to be a “prevailing party” under EAJA, however, a party need not prevail on all of its claims. It must only “succeed on any significant issue in litigation which achieves some of the benefit the parties sought in bringing suit.” Nadeau, 581 F.2d at 278-79. The district court did not abuse its discretion in finding that plaintiffs met this standard. Not only did the individual plaintiffs obtain a remand— and, ultimately, their benefits — as a result of the litigation, but some 10,000 members of the plaintiff class were assured that their disability status would be considered or reconsidered under the de minimis severity test. The Secretary concedes that this was a “significant victory.” Appellant’s Reply Brief at 4. In view of these notable successes, neither a final victory on their claim of facial invalidity nor a judicial finding of invalidity as applied was necessary to make the plaintiffs “prevailing parties.”
Finally, the Secretary concedes the plaintiffs’ success on the “combination of impairments” issue, but argues that this was too slight a victory to make plaintiffs “prevailing parties.” Because we accept the district court’s finding that plaintiffs prevailed in their more central effort — to obtain reconsideration of their Step 2 denials under a de minimis standard — we find this argument inapposite.
V. SUBSTANTIAL JUSTIFICATION
In order to justify an EAJA award, the district court must find not only that the plaintiffs were “prevailing parties,” but also that the government’s position was not “substantially justified.” The Supreme Court has explained that for a government position to be “substantially justified,” it must have “a reasonable basis in law and fact” and be “justified to a degree that could satisfy a reasonable person.” Underwood, 108 S.Ct. at 2550 and 2550 n. 2. The burden of showing such justification is upon the government. See United States v. Yoffe, 775 F.2d 447, 450 (1st Cir.1985).
In the first years after the enactment of EAJA, courts were split about which “government positions” had to be “substantially justified” in order to preclude an EAJA award. Some courts held that the underlying government action needed to be justified, while others held that only the government’s litigating position needed to be justified. Congress resolved this controversy in the 1985 amendment to EAJA. The statute, as amended, makes clear that [1476]*1476“ ‘position of the United States’ means, in addition to the position taken by the United States in the civil action, the action or failure to act by the agency upon which the civil action is based.” 28 U.S.C. § 2412(d)(2)(D) (emphasis added). See also H.R.Rep. No. 120, Part I, 99th Cong., 1st Sess. 9, reprinted in 1985 U.S.Code Cong. & Admin.News 132, 137 (explaining that some courts had construed “position of the United States” in an overly narrow way “which has helped the Federal Government escape liability for awards”); Sierra Club v. Secretary of the Army, 820 F.2d 513, 517 (1st Cir.1987) (“Despite earlier uncertainty, it is by now well settled that the ‘position’ which must be justified comprises ‘both the position of the agency and the litigation position of the government’ ” (quoting Yoffe, 775 F.2d at 449)). In the present case we can concede that many of the government’s litigating positions were reasonable and, hence, “substantially justified.” The central question facing us, however, is whether the underlying agency action was reasonable.
The district court found that the government did not meet its burden of showing that its application of the Step 2 severity regulation, during the period that preceded the adoption of SSR 85-28, had been reasonable.7 This was the period when the four named plaintiffs’ claims were initially rejected by HHS and when many of the class members would have had their claims reviewed by the agency. The court noted that
the government has offered little or no justification for the Secretary’s application of the severity test, even in the face of substantial evidence — the statistics as to the number of claims denied, the rulings of many other courts and, perhaps most significantly, the Secretary’s own Ruling 85-28 — indicating that the regulations had been misapplied.
McDonald v. Bowen, 693 F.Supp. at 1304.
In making this determination, the district court did not abuse its discretion. To determine what would have been a justifiable application of the Step 2 severity regulation, one must compare the regulation with the Act it was supposed to implement. The Social Security Act provides that a person is disabled if his impairment or impairments prevent him from doing gainful work, “considering his age, education, and work experience.” 42 U.S.C. § 423(d)(2)(A) (emphasis added). The Step 2 severity regulation, introduced in 1978 as part of a new five-step determination process, screened out claimants who did not have a “severe” impairment, without making reference to their age, education, or work experience. The Supreme Court subsequently upheld the regulation to the extent that it merely screened out, at an early stage, “those claimants whose medical impairments are so slight that it is unlikely they would be found to be disabled even if their age, education and experience were taken into account.” Yuckert, 482 U.S. at 153, 107 S.Ct. at 2297. But the Court did not state or imply that it would approve the application of the regulation in a way that went beyond such a de minimis screening.
The present record is replete with indications that the regulation did become “in practice, more than a de minimis screening device.” McDonald I, 795 F.2d at 1124. As we noted in McDonald I, the available statistics suggested that the severity regulation led to a large increase in the number of applicants denied benefits on the basis of medical factors alone. See id. at 1124 (citing a House of Representatives report cited by Baeder v. Heckler, 768 F.2d 547, 552 (3d Cir.1985)). Justice O’Connor, in her concurring opinion in Yuckert, surveyed the empirical evidence and came to a similar conclusion:
Before the step two regulations were promulgated approximately 8% of all claimants were denied benefits at the “not severe” stage of the administrative process; afterwards approximately 40% of all claims were denied at this stage.... As the lower federal courts [1477]*1477have enjoined use of step two and imposed narrowing constructions, the step two denial rate has fallen to about 25%.... [Respondent's evidence suggests that step two has been applied systematically in a manner inconsistent with the statute.
Yuckert, 482 U.S. at 157, 107 S.Ct. at 2299 (O’Connor, J., concurring) (citations omitted). The government’s application of a regulation in a manner dramatically inconsistent with its authorizing statute cannot be “substantially justified.”
We also note the consistent view of the federal courts of appeals that the severity regulation had been misapplied. The Supreme Court in Yuckert (like this court in McDonald I) did not find it necessary to pass on whether the severity regulation had been properly applied. See Yuckert, 482 U.S. at 154 n. 12, 107 S.Ct. at 2297 n. 12. But as Justice O’Connor noted in her concurring opinion,
Despite the heavy deference ordinarily paid to the Secretary’s promulgation and application of his regulations, ... all 11 regional Federal Courts of Appeals have either enjoined the Secretary’s use of the step two regulation or imposed a narrowing construction upon it. The frustration expressed by these courts in dealing with the Secretary’s application of step two in particular cases is substantial....
Id. at 156-57, 107 S.Ct. at 2298-99 (O’Con-nor, J., concurring) (citation and footnotes omitted). The fact that one or more courts have invalidated an agency’s policy does not necessarily indicate that the policy was unreasonable. See Underwood, 108 S.Ct. at 2552. Nevertheless, “a string of losses can be indicative.” Id. In litigation, as in sports, a 0-11 record suggests that something is amiss.
Given the statistical evidence and the extensive judicial criticism (not to mention the agency’s own eleventh hour issuance of SSR 85-28 as an apparent corrective measure, see McDonald I at 1124), the agency’s earlier application of the severity regulation was, at very least, under a cloud. In light of this, the agency’s burden to justify its position was not a perfunctory one. Yet the district court supportably found that “the government has offered little or no justification for the Secretary’s [prior] application of the severity test.” McDonald, 693 F.Supp. at 1304. Even on this appeal the Secretary does not attempt to justify his application of the severity regulation prior to the adoption of SSR 85-28. We think, therefore, that the district court did not err in determining that HHS had not justified its earlier application of the severity test.
Nevertheless, the Secretary contends that we should find that the government has met its burden of establishing that its position was substantially justified. First, the Secretary reminds us that in McDonald II this court commented that it was faced with a “close and perplexing” question. McDonald II at 1089. But we were perplexed chiefly by the jurisdictional issue raised by the Secretary in McDonald II. The Secretary’s contention goes to the reasonableness of the government’s litigating position in McDonald II — which we can concede — not to whether the HHS’s prior application of the severity regulation had been reasonable.8
Second, the Secretary contends that because neither the district court nor this court made a specific finding that the severity regulation had been invalidly applied, it was “bizarre” for the district court to find that the Secretary’s position had not been justified. But, as we have already held, the court below supportably found that plaintiffs had in effect prevailed on their challenge to the severity regulation as applied. McDonald, 693 F.Supp. at 1304. Given this, there is nothing bizarre about the district court’s further finding that HHS had applied the regulation unreasonably.
Finally, the Secretary argues that we should follow Baeder v. Heckler, 826 F.2d [1478]*14781345 (3d Cir.1987) and Mattson v. Bowen, 824 F.2d 655 (8th Cir.1987), in which the courts foreclosed an EAJA award to prevailing plaintiffs on the ground that the severity regulation was substantially justified. As far as we can tell, however, the attorneys’ fees denied in Baeder and Matt-son were for legal work solely in bringing facial challenges to the severity regulation.9 Although the plaintiffs prevailed in these challenges the Supreme Court’s subsequent decision in Yuckert made clear that the government’s position had been reasonable. The present plaintiffs, in contrast, obtained significant relief in their challenge to the severity regulation as formerly applied. Nothing in Yuckert indicated approval of the Secretary’s prior application of the severity regulation.
The district court also found that the Secretary failed to establish that his position on the pre-1984 combination of impairments issue was substantially justified. McDonald, 693 F.Supp. at 1302-03. On appeal, the Secretary insists that the HHS’s pre-1984 policy of not considering the combined effects of non-severe impairments was in fact a reasonable interpretation of the Act. This argument, relying on a combination of linguistic analysis (the Social Security amendments of 1954 referred to “impairment” in the singular) and legislative exegesis (when Congress amended the Act in 1984, it did not declare HHS’s past practice to have been illegal), is creative but unpersuasive. Not only the district court and this court but four other circuit courts of appeals have rejected the Secretary’s position. See McDonald I, 795 F.2d at 1127 (citing cases). As we noted in McDonald I,
It seems simply a matter of common sense that various physical, mental, and psychological deficits, each non-severe in and of itself, might in combination, in some cases, make it impossible for a claimant to work.
McDonald I, 795 F.2d at 1127.
We thus agree with the district court that the government has not met its burden of showing that either of the challenged policies was a reasonable interpretation of the law.
VI. THE SCOPE OF PLAINTIFFS’ RECOVERY OF ATTORNEYS’ FEES
Having accepted the district court’s determination that the plaintiffs were prevailing parties and that the challenged agency policies were not substantially justified, we turn to the scope of the attorneys’ fees awarded to plaintiffs. Plaintiffs did not succeed on all of their legal claims. Specifically, they failed in their effort— which persisted even after the issuance of SSR 85-28 — to invalidate the severity regulation as written. Recognizing this, the district court confronted the question of what portion of plaintiffs’ attorneys’ time charges should be paid by the government under EAJA.
In determining the proper scope of plaintiffs’ fee recovery, the district court relied upon the Supreme Court’s approach in Hensley v. Eckerhart, 461 U.S. 424, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983). In applying the Civil Rights Attorney’s Fees Awards Act, 42 U.S.C. § 1988, the Hensley Court noted that
In [some] cases the plaintiff’s claims for relief will involve a common core of facts or will be based on related legal theories. Much of counsel’s time will be devoted generally to the litigation as a whole, making it difficult to divide the hours expended on a claim-by-claim basis. Such a lawsuit cannot be viewed as a series of discrete claims. Instead the district court should focus on the significance of the overall relief obtained by the plaintiff in relation to the hours reasonably expended on the litigation.
Id. 461 U.S. at 435, 103 S.Ct. at 1940.
Here the district court found that much of plaintiffs’ counsel’s work was relevant to both the “facial” and the “as applied” challenges to the severity regulation:
[1479]*1479At each step of the litigation, the plaintiffs asserted that the severity regulations, both as written and as applied, had resulted in the denial of disability benefits to claimants entitled to those benefits under the Social Security statute. Work on these two different challenges necessarily overlapped, and could thus be attributed to either theory.
The plaintiffs’ successful challenge to the severity regulations as applied raised largely the same issues as their unsuccessful attack on the facial validity of these regulations.
McDonald, 693 F.Supp. at 1304-05. The court found that as a result of their challenge to the severity regulation, as well as to the combination of impairments regulation, “the plaintiffs essentially succeeded in winning the relief they had sought.” Id. at 1305. Accordingly, the court awarded to the plaintiffs fees for all of the billable hours claimed.10
In Sierra Club v. Secretary of the Army, 820 F.2d at 520 n. 6, we left open the question of the applicability of section 1988 jurisprudence to EAJA cases involving “mixed bags of claims — some successful, some not.” We now confront that question and conclude that the district court’s strategy of looking to Hensley was appropriate. As discussed earlier, supra at 1474, the standards developed under section 1988 for defining a “prevailing party” are applicable to EAJA cases. Likewise the section 1988 ease law concerning the scope of recovery of attorneys’ fees should also be applicable to EAJA cases. See Hensley at 461 U.S. at 433 n. 7, 103 S.Ct. at 1939 n. 7 (noting that Hensley’s standards “are generally applicable in all cases in which Congress has authorized an award of fees to a ‘prevailing party’ ”).
Although we approve the district court’s reliance on Hensley, we believe that in two respects the court applied the Hensley factors in an insufficiently discriminating and overly generous way.
First, we are dubious about the court’s conclusion that plaintiffs “essentially succeeded in winning the relief they had sought.” McDonald, 693 F.Supp. at 1705. The plaintiffs did succeed in obtaining reconsideration of their claims on the basis of a de minimis severity test, and this — together with the disability awards that at least the named plaintiffs obtained after remand — was undoubtedly a major portion of what plaintiffs sought.11 But plaintiffs were reaching for more than that. They also sought to invalidate the severity regulation on its face, and this prize they did not win. Plaintiffs persisted, moreover, in their facial challenge even after HHS had issued SSR 85-28, which gave the severity regulation a de minimis interpretation. If all that plaintiffs wanted was reconsideration of their claims under a de minimis standard, they would have had no reason to pursue the facial challenge after SSR 85-28 was issued. In light of Hensley’s instruction that the court “should focus on the significance of the overall relief obtained in relation to the hours reasonably expended on the litigation,” 461 U.S. at 435, 103 S.Ct. at 1940, the district court should have reduced the plaintiffs’ attorneys’ fee.
Second, as the district court correctly noted, “Because the EAJA was designed to facilitate challenges to unreasonable government action, ‘it would contravene the purposes of the Act to require the government to bear the expense of defending even its reasonable positions.’ ” McDonald, 693 F.Supp. at 1304 (quoting Matthews v. United States, 713 F.2d 677, 684 (11th Cir.1983)) (emphasis in original). Given that the Supreme Court (in Yuckert) has made clear that the government’s posi[1480]*1480tion with respect to the facial validity of the severity regulation was reasonable, the district court should have reduced the fee award to reflect the extent to which the plaintiffs’ legal efforts were devoted to the facial challenge.
We conclude that the district court abused its discretion in failing to reduce the scope of plaintiffs’ fee recovery. Plaintiffs’ counsel did not break down their itemized account of their time to distinguish “facial challenge hours” from “ ‘as applied’ challenge hours.” 12 Given the district court’s finding, which we accept, that the work on the two challenges overlapped to a large degree, we do not fault counsel for not providing such a breakdown. Nevertheless, the inability to neatly label hours as “facial” or “as applied” does not mean that the court is powerless to make an appropriate adjustment. The district court could have—and should have—equitably reduced the overall award to reflect the plaintiffs’ less than complete success and the extent to which their legal work was aimed at contesting the government’s reasonable position. See Hensley, 461 U.S. at 436-37, 103 S.Ct. at 1941 (“The district court may attempt to identify specific hours that should be eliminated, or it may simply reduce the award to account for the limited success”).
Rather than remand this matter to the district court and thus further prolong this fee litigation, we make our own equitable judgment and direct that the district court’s fee award be reduced by 20 percent. The 20 percent figure is necessarily somewhat arbitrary. Nevertheless, it reflects our judgment on consideration of counsel’s fee submission that at least 20 percent of the hours billed could have been avoided if plaintiffs had not chosen to pursue their ill-fated challenge to the severity regulation on its face.
VII. ATTORNEYS’ FEES FOR WORK ON THE FEE LITIGATION
The district court included in its award fees for the work performed by plaintiffs’ attorneys on the fee application itself. McDonald, 693 F.Supp. at 1305-06 n. 4. We agree with the district court that such an award is appropriate.
As a general rule, the time reasonably devoted by attorneys to successful fee applications can be reimbursed under fee-shifting statutes, such as section 1988. See, e.g., Cinciarelli v. Reagan, 729 F.2d 801, 809 (D.C.Cir.1984); Prandini v. National Tea Co., 585 F.2d 47, 53-54 (3d Cir.1978) (citing cases); Souza v. Southworth, 564 F.2d 609, 614 (1st Cir.1977). Such a result seems consistent with the purpose of such statutes to ease litigants’ path in vindicating their legal rights. “To hold otherwise would permit a deep pocket losing party to dissipate the incentive provided by an award through recalcitrance and automatic appeals.” Souza, 564 F.2d at 614. Because of EAJA’s focus on the “substantial justification” of the government’s position, however, it can be argued that this general rule is not necessarily applicable to EAJA cases. Thus, one might read EAJA as providing that if the government’s position with respect to the fee litigation was substantially justified, plaintiffs should not receive attorneys’ fees for this portion of the litigation. Some courts have adopted this line of reasoning. See Lee v. Johnson, 799 F.2d 31, 39-40 (3d Cir.1986); Cornella v. Schweiker, 741 F.2d 170 (8th Cir.1984).
We are persuaded, however, to adopt the per se fee-shifting rule adopted by the Second Circuit:
[Wjhere the government’s underlying position is not substantially justified, plaintiff is entitled under the EAJA to recover all attorney’s fees and expenses reasonably incurred in connection with the vindication of his rights, including those related to litigation over fees, and any appeal.
Trichilo v. Secretary of Health and Human Services, 832 F.2d 743, 745 (2d Cir.1987). This rule is recommended by considerations of judicial economy. If courts had to assess the “substantial justification” of [1481]*1481each government decision to oppose a fee application, we could be faced with “an infinite regression of EAJA litigation.” Cinciarelli, 729 F.2d at 810. By adopting such a per se rule we can best conform to the Underwood Court’s direction that a “ ‘request for attorney’s fees should not result in a second major litigation.’ ” Underwood, 108 S.Ct. at 2549 (quoting Hensley, 461 U.S. at 437, 103 S.Ct. at 1941).13 In addition, we agree with the district court that “the remedial purpose of the EAJA is best served by awarding fees against the government for all phases of litigation made necessary by its unreasonable actions.” McDonald, 693 F.Supp. at 1306 n. 4. See also Trichilo v. Secretary of Health and Human Services, 823 F.2d 702, 707 (2d Cir.1987) (“Since the purpose of the EAJA is to remove counsel fees as an impediment to challenging unreasonable and unjustified governmental actions, where a governmental action has been shown to have been unjustified, there should be as little disincentive for plaintiffs to obtain attorney’s fees as there is for them to challenge the action itself”). This viewpoint is further supported by the 1985 amendment to EAJA, which made clear that “the government’s position” under EAJA refers to the underlying agency actions as well as to the government’s litigating position. 28 U.S.C. § 2412(d)(2)(D).
Applying this per se fee-shifting rule, we affirm the district court’s award of attorneys’ fees for work on the EAJA petition and litigation. Plaintiffs may also apply to the district court for attorneys’ fees reasonably incurred in connection with the present appeal.14
VIII. CONCLUSION
In summary, we hold that the district court did not abuse its discretion in determining that plaintiffs were “prevailing parties” and that the underlying agency position was not shown to be “substantially justified.” But we direct that the fee awarded by the district court be reduced by 20 percent, to reflect the extent to which plaintiffs’ legal work was devoted to the facial challenge to the severity regulation. Finally, we affirm the district court’s award of attorney fees for work on the fee litigation itself.
Affirmed in part and remanded for further proceedings in accord with the foregoing instructions.