Opinion for the Court filed by Circuit Judge SENTELLE.
Dissenting Opinion filed by Circuit Judge D.H. GINSBURG. SENTELLE, Circuit Judge:Appellant Secretary of Health and Human Services (“the Secretary” or “appellant”) appeals from two District Court orders allowing attorneys’ fees against the Secretary under 28 U.S.C. § 2412(b) on findings that appellant acted in “bad faith.” *399Because we hold that the District Court’s finding of bad faith on the part of appellant is not clearly erroneous, we affirm.
I. Background
The attorneys’ fees issue arises in the context of a lawsuit brought by the American Hospital Association and eleven additional hospital associations (collectively “AHA”), on behalf of their member hospitals, challenging the Department of Health and Human Services’s (“HHS” or “appellee”) Medicare secondary payer regulations.
In 1980, Congress amended 42 U.S.C. § 1395y(b)(l) to exclude from Medicare coverage services that had been paid for or could reasonably be expected to be paid for by liability insurance. Under § 1395y(b)(l) Medicare was essentially a secondary payer; hospitals were required to pursue payment from liability insurance carriers for services when they reasonably could expect those payments to be made promptly before Medicare payments would be made. In 1988, HHS issued a revised Manual section, see Medicare Intermediary Manual § 3419 (1988), and proposed regulations, see Medicare as Secondary Payer and Recovery Against Third Parties, 53 Fed.Reg. 22,335 (1988) (to be codified as 42 C.F.R. pts. 405, 411 & 489) (proposed June 15, 1988), prohibiting hospitals from pursuing payment from liability insurers under any circumstances and requiring hospitals to accept payment from Medicare. The Secretary pressed for the change because “the hospital’s collection of larger sums from insurers than they could collect from Medicare ... allow[ed] providers to invade sums that would otherwise be available to compensate the beneficiary for other items of damages, such as lost earnings or pain and suffering.” Brief for the Appellants at 5.
Shortly after HHS’s notice of proposed rulemaking, AHA filed a complaint seeking declaratory and injunctive relief from HHS’s revised policy. After hearing, and in lieu of a preliminary injunction, the Secretary of HHS entered into a stipulation with AHA specifying that, pending the outcome of the suit, AHA’s member hospitals could continue to bill liability insurers directly for services provided and could file liens against recoveries by Medicare beneficiaries from third-party liability insurers, so long as any such sums collected were placed in escrow. Under a Joint Stipulation and Order (“Stipulation”), HHS agreed not to
terminate, institute termination proceedings, or threaten to terminate the Medicare provider agreement of any member hospital, or impose any other sanction ... on grounds that the member hospital has billed willing liability insurance carriers ... under the circumstances as described in 42 U.S.C. § 1395y(b)(l) or any manual provisions, directives, instructions, or rules or regulations, existing or proposed, that interpret or implement 42 U.S.C. § 1395y(b)(l).
Stipulation, Civ. No. 88-2027 at 2, 3 (Aug. 24, 1988).
Despite the existence of the Stipulation, on October 11, 1989, the Secretary published regulations prohibiting hospitals from billing directly liability insurers or placing liens on Medicare beneficiaries’ recoveries from third-party insurers. See Medicare as Secondary Payer and Medicare Recovery Against Third Parties, 54 Fed.Reg. 41,716 (1989) (to be codified at 42 C.F.R. pts. 405, 411, 412, and 489) (“Final Rule”). In the Final Rule, which was to become effective November 13, 1989, the Secretary made an express exemption of hospitals involved in a lawsuit against HHS in Oregon, but no exemption for AHA hospitals.
One week after publication of the Final Rule, AHA filed a Second Motion for Preliminary Injunction requesting that the District Court restrain HHS from enforcing the regulation against its member hospitals. The Secretary responded by filing a Renewed Motion to Dismiss and Opposition to the Second Motion for Preliminary Injunction, arguing, inter alia, that the Final Rule was never meant to supersede the Stipulation. On the same day, October 26, 1989, the Secretary issued a directive to his regional administrators requiring them to send letters of clarification to intermediaries (responsible for the initial process of *400Medicare claims) stating that the regulation would not apply to AHA hospitals.
On November 9, 1989, the District Court ruled that HHS’s enforcement of certain provisions of the regulation would be contrary to the Stipulation, enjoined HHS from enforcing the Final Rule against AHA’s member hospitals, and required HHS to publish the order granting the injunction along with a copy of the Stipulation in the Federal Register as an exemption to the Final Rule. Order, Civ. No. 88-2027 at 2-3 (Nov. 9, 1989). The court also awarded AHA costs, expenses, and attorneys’ fees incurred in preparing and filing the second motion for an injunction. Id. at 3-4.
On May 24, 1990, the District Court denied HHS’s motion to reconsider the award of fees, ruling that HHS’s publication of the Final Rule without an exception for AHA violated the Stipulation and was done “willfully, and in bad faith.” Memorandum Opinion and Order, Civ. No. 88-2027 at 3 (May 24, 1990). The court awarded fees in the amount of $16,914.77 for the preparation and filing of the second motion for an injunction. The court also authorized AHA to apply for fees incurred in preparation of its reply memorandum in response to HHS’s motion to reconsider the fee award. Based on that reply, the court awarded $3,432.27 in additional fees on August 7, 1990. Memorandum Opinion and Order, Civ. No. 80-2027 (Aug. 7, 1990).
HHS appeals from the May 24th and August 7th awards of attorneys’ fees. HHS does not appeal the final judgment in the underlying litigation.
II. Analysis
This case involves an atypical application of the Equal Access to Justice Act (“EAJA”). That Act waived the sovereign immunity of the United States against attorneys’ fees in two distinct manners. Most EAJA litigation arises under 28 U.S.C. § 2412(d)(1)(A), which provides for the award of fees against the United States in most types of civil litigation “unless the court finds that the position of the United States was substantially justified or that special circumstances make an award unjust.” See, e.g., Pierce v. Underwood, 487 U.S. 552, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988). However, that section erects certain threshold requirements for the party seeking counsel fees, one of which is that its “net worth ... not exceed $7,000,000 at the time the civil action was filed and [that it] had not more than 500 employees.” 28 U.S.C. § 2412(d)(2)(B). Appellees, apparently not meeting this or other requirements, do not assert eligibility for a fee award under § 2412(d).
»[1] There is, however, a lesser used waiver of sovereign immunity against attorneys’ fees in 28 U.S.C. § 2412(b). That section makes “[t]he United States ... liable for [attorneys’] fees and expenses to the same extent that any other party would be liable under the common law or under the terms of any statute which specifically provides for such an award.” In this case, the District Court applied § 2412(b) and looked to the common law for the underlying liability. Under the American rule, common law as a general matter precludes the award of attorneys’ fees from the losing party to the prevailing party. A narrow exception has developed where the losing party has acted in “bad faith.” See F.D. Rich Co. v. United States, 417 U.S. 116, 129, 94 S.Ct. 2157, 2165, 40 L.Ed.2d 703 (1974) (“[Attorneys’ fees may be awarded to a successful party when his opponent has acted in bad faith, vexatiously, wantonly, or for oppressive reasons”); Alyeska Pipeline Serv. Co. v. Wilderness Soc’y, 421 U.S. 240, 247, 95 S.Ct. 1612, 1616, 44 L.Ed.2d 141 (1975) (“In the United States, the prevailing litigant is ordinarily not entitled to collect a reasonable attorneys’ fee from the loser.”); Lipsig v. Nat’l Student Mktg. Corp., 663 F.2d 178, 180 (D.C.Cir.1980) (noting “narrowly defined” bad faith exception to the American rule).
Bad faith can support an award of attorneys’ fees in circumstances where the bad faith (1) occurred in connection with the litigation, or (2) was an aspect of the conduct giving rise to the lawsuit. Nepera Chem., Inc. v. Sea-Land Serv., Inc., 794 F.2d 688, 701 (D.C.Cir.1986). Examples of *401the first include the filing of a frivolous complaint or meritless motion, see Lipsig, 663 F.2d at 182, or discovery-related misconduct, Fritz v. Honda Motor Co., 818 F.2d 924 (D.C.Cir.1987).
Bad faith in conduct giving rise to the lawsuit may be found where “a party, confronted with a clear statutory or judicially-imposed duty towards another, is so recalcitrant in performing that duty that the injured party is forced to undertake otherwise unnecessary litigation to vindicate plain legal rights.” Fitzgerald v. Hampton, 545 F.Supp. 53, 57 (D.D.C.1982). See also American Employers Ins. Co. v. American Sec. Bank, 747 F.2d 1493, 1502 (D.C.Cir.1984) (“[Ejxception to the American rule ... allows an award of attorneys’ fees when the party has been the victim of unwarranted, oppressive, or vexatious conduct on the part of his opponent and has been forced to sue to enforce a plain legal right.”).
Here, the District Court found that publication of the Final Rule directly contradicting the Stipulation was done in bad faith. As this conduct which gave rise to AHA’s motion is analogous to conduct giving rise to litigation, the court honored AHA’s prayer for counsel fees. The Stipulation had preserved effectively the status quo, allowing member hospitals to bill directly liability insurers for services to Medicare beneficiaries and to file liens against beneficiaries’ recoveries from third-party insurers. Under the Stipulation, HHS agreed not to sanction member hospitals following its terms. The Final Rule, however, expressly prohibited the billing practices preserved by the Stipulation and on its face threatened to terminate member hospitals from participation in the Medicare program should such practices occur. The District Court found that the promulgation of the Final Rule was “at best, ... calculated to confuse [AHA’s] member hospitals,” and “[a]t worst, ... was an attempt to chill. [AHA] member hospitals’ assertion of their rights as they had been defined by this court’s Stipulation and Order.” Memorandum Opinion and Order, supra, at 399. Because the Stipulation had the force of law and because the District Court found that publication of the Final Rule contradicted the Stipulation, the District Court recognized that HHS effectively forced AHA to sue to enforce the plain legal rights defined by the Stipulation.
The District Court’s finding of bad faith is likewise supported by the finding that HHS did not exempt expressly AHA member hospitals from operation of the Final Rule. The absence of such an exemption is especially significant to the bad faith determination because, on the face of the published rule, HHS exempted certain hospitals in Oregon that had challenged the new HHS regulations in separate litigation. HHS argued that the failure to include a similar exemption for AHA member hospitals was a mere bureaucratic oversight. The District Court rejected this contention, finding that HHS was no less aware of this case than of the Oregon litigation and that the failure to include protection for these plaintiffs was not the result of oversight. Id. at 399. Given the state of the evidence before the District Court, we cannot say that its determination of bad faith was “clearly erroneous.” There was evidence to support that determination, and we are not “ ‘on the entire evidence ... left with the definite and firm conviction that a mistake has been committed.’ ” Anderson v. Bessemer City, 470 U.S. 564, 573, 105 S.Ct. 1504, 1511, 84 L.Ed.2d 518 (1985) (quoting United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 542, 92 L.Ed. 746 (1948)).
Our language in the last two sentences is, of course, appropriate to a review of a finding of fact, not a conclusion of law. We use that language advisedly. The Secretary argues that we should review the District Court’s determination of bad faith de novo as a conclusion of law rather than under the more deferential standard of review afforded findings of fact. (“Findings of fact ... shall not be set aside unless clearly erroneous____” Fed.R.Civ.P. 52(a)). The Secretary offers in support of this position our decision in Nepera Chemical, Inc. v. Sea-Land Service, Inc., supra, wherein we stated that “the standard for a *402finding of bad faith is stringent.” Nepera Chem., 794 F.2d at 702. That language, however, does not establish or even deal with the question of whether the determination is one of law or fact; rather, it addresses the degree of certainty with which the maker of the decision must act, whichever type decision is involved.
To the same effect is Lipsig v. National Student Marketing Corp., supra, cited by the Secretary in further support of the proposition. Indeed, both parties cite Lipsig for opposite conclusions. As with Nepera Chemical, the Secretary cites Lipsig’s statement that “ ‘the standards for bad faith are necessarily stringent,’ ” as support for the proposition that the bad faith determination is one of law. Id. at 180 (quoting Adams v. Carlson, 521 F.2d 168, 170 (7th Cir.1975)). In contrast, AHA cites Lipsig for its analysis wherein this Court reviewed the factual findings supporting a determination of bad faith under a clearly erroneous standard. Id. at 181-82. In AHA’s view, the case stands for the proposition that “a district court’s finding of bad faith, as well as the subsidiary findings that support it, are factual findings that may be reversed only if ‘clearly erroneous.’ ” Brief for the Appellees at 15 (citing Lipsig, 663 F.2d at 181).
Unfortunately, the Lipsig opinion is not dispositive of either position. If it were, our choice would be an easy one since we would be bound by circuit law. However, the Lipsig opinion determined only that the standard was a stringent one, whether of fact or of law, and that the subsidiary findings were reviewed under a “clearly erroneous” standard. The court functioned in silence as to the standard applicable to the second tier determination of bad faith. Nonetheless, Lipsig is instructive to our decision. After having stated, “we cannot say that the court’s findings are clearly erroneous,” Lipsig, 663 F.2d at 181 (footnote omitted), the Lipsig panel went on to state that it could not say that “the court exceeded the bounds of its discretion in holding that [the subsidiary findings] called for an award of compensatory attorneys’ fees.” Id. at 181-82. The panel’s reference to the District Court’s discretion makes it appear that the panel did not shift review standard gears between making its subsidiary findings and its finding of bad faith based thereon.
Though we and the parties have mined the decisional vein of this Court’s past opinions thoroughly, we have not found one that answers the question any more squarely than does Lipsig. As noted, that decision more closely supports appellee’s position than the Secretary’s. Moreover, another case cited by both sides, ALPO Petfoods, Inc. v. Ralston Purina Co., 913 F.2d 958 (D.C.Cir.1990), also supports that position. The Secretary relies on our discussion in that case to the effect that the District Court’s factual findings were inadequate to meet applicable legal standards of bad faith — in particular, the need to show deliberate intent to harm. While the Secretary’s presentation is accurate, our analysis in ALPO is not inconsistent with a clearly erroneous review. Appellee argues more convincingly from our language expressly reversing the District Court’s award of attorneys’ fees to plaintiff. We did so noting that “neither the court’s opinion nor its later statements support the award with a finding of willfulness or bad faith, and since we have decided that any such finding would be clearly erroneous ... we reverse the district court’s decision to award attorneys’ fees to ALPO.” Id. at 971. While this does not amount to an express holding on our part that the question is to be reviewed as a finding of fact, we would hardly have used the phrase “clearly erroneous” unless that was the standard for the question before us.
Thus, though circuit law may not be so fixed as to compel us to hold that the question before us is one of fact requiring a clearly erroneous standard of review, it at least impels us in that direction. In determining finally to follow the path toward which Lipsig and ALPO have directed us, we look to the rationale underlying the clearly erroneous standard established by Federal Rule of Civil Procedure 52(a). When the question requires a court to determine credibility or to compare the weights of evidentiary items, an appellate *403court must remember two humbling factors: (1) “the superiority of the trial judge’s position to make determinations of credibility,” and (2) the principle that “[t]he trial judge’s major role is the determination of fact, and with experience in fulfilling that role comes expertise.” Anderson, supra, 470 U.S. at 574, 105 S.Ct. at 1511. Therefore, we conclude that the question of bad faith in the context of the common law exception to the American rule on counsel fees — involving as it does credibility determinations and evidentiary weighing as opposed to the application of rules of the law to matters of fact (either conceded or already determined) — is one of fact requiring a clearly erroneous standard of review, not one of law requiring de novo review. It is certainly possible in a specific case that an evidentiary base might not support findings of fact consistent with a finding of bad faith as a matter of law or that subsidiary findings on such evidence might not support an ultimate finding of bad faith. See, e.g., ALPO Pet Foods, supra, 913 F.2d at 975 (reversing district court’s finding of bad faith where it was not supported within the court’s opinion). But the same is true with findings of fact in general, or else Rule 52(a) would not provide for a “clearly erroneous” standard of review.
While not essential to our decision, we note that other circuits have reached the same conclusion when considering the question directly. In Brown v. Sullivan, 916 F.2d 492 (9th Cir.1990), the Ninth Circuit stated that “[a] district court’s finding regarding a party’s bad faith is reviewed under the clearly erroneous standard.” Id. at 495. Likewise, the Third Circuit in Ford v. Temple Hospital, 790 F.2d 342 (3d Cir. 1986), found it “established that a district court’s finding of bad faith or the absence of bad faith in a particular case is a factual determination and may be reversed only if it is clearly erroneous.” Id. at 347. We find no circuits to the contrary. We further note that our conclusion is consistent with the Supreme Court’s teaching in, among other decisions, Pullman Standard v. Swint, 456 U.S. 273, 102 S.Ct. 1781, 72 L.Ed.2d 66 (1982), that the clearly erroneous standard of review in Rule 52(a) “does not divide facts into categories; in particular it does not divide findings of facts into those that deal with ‘ultimate’ and those that deal with ‘subsidiary’ facts.” Id. at 287, 102 S.Ct. at 1789. Since it appears to us for reasons already set forth that this is a finding of ultimate fact not a conclusion of law, the standard of Rule 52(a) prevails and the District Court’s finding will stand.
We are not inadvertent to the Secretary’s argument that the District Court’s finding of willfulness is unsupportable because the Secretary’s promulgation of the Final Rule did not afford the Secretary a litigation advantage. However, litigation advantage is not the sine qua non of willfulness. The District Court’s findings, as noted above, that the publication was “at best, ... calculated to confuse plaintiffs’ member hospitals [and,] at worst, ... an attempt to chill plaintiffs’ ... assertion of their rights”; that the defendant “did not overlook the effect of the Oregon litigation” and “could have been no less aware of the 1988 Stipulation and Order of this court”; and that “publication of the Final Rule without carving out an exception for the plaintiffs here was obviously not an oversight” sufficiently support the ultimate finding of bad faith. These findings are consistent with the standard that permits an award of attorneys’ fees for bad faith where the party receiving the award “has been the victim of unwarranted, oppressive, or vexatious conduct on the part of his opponent and has been forced to sue to enforce a plain legal right.” American Employers Ins., supra, 747 F.2d at 1502. This principle is no less applicable where the action the victim is compelled to take is the filing of a motion rather than the filing of the law suit ab initio.
III. Conclusion
For the reasons set forth above, we conclude that the District Court’s finding of bad faith is reviewable under a clearly erroneous standard. We further conclude that the finding is not clearly erroneous. Therefore, the District Court’s award of counsel fees is
Affirmed.