concurring.
Although I join in the court’s opinion, I find this case far more difficult than its tone would suggest. To hold final an award of counsel fees after a partial merits judgment itself certified as final under Fed. R.Civ.P. 54(b), as we do, is to presuppose that an appropriate fee can and should be determined at that time. I find this assumption somewhat problematic, and indeed in some tension with the Supreme Court’s prescribed method for determining an appropriate counsel fee award. I write separately to identify how today’s result will complicate the counsel fee analysis and to explain why I believe that we reach the right result nonetheless.
I.
In Hensley v. Eckerhart, 461 U.S. 424, 103 S.Ct. 1933, 76 L.Ed.2d 40 (1983), the Supreme Court established the basic methodology for calculating an appropriate award of statutory attorneys’ fees. “The most useful starting point for determining the amount of a reasonable fee is the number of hours reasonably expended on the litigation multiplied by a reasonable hourly rate,” id. at 433, 103 S.Ct. at 1939—the so-called “lodestar” amount. However, hours spent litigating unsuccessful claims unrelated to the ones on which the plaintiff has prevailed must then be identified and excluded from the lodestar. See id. at 434-35, 103 S.Ct. at 1939-40. If, on the other hand, the unsuccessful claims are sufficiently interrelated with the successful ones — i.e. they “involve a common core of facts or [are] based on related legal theories,” id. at 435, 103 S.Ct. at 1940—courts should not attempt to identify specific hours spent on those claims and to exclude them from the lodestar. “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.
Hensley thus teaches that the requisite comparison between hours expended and relief obtained cannot be done on a claim-by-claim basis for claims that “involve a common core of facts or [are] based on related legal theories.” Thus, a plaintiff who prevailed on only one of several interrelated claims and won essentially complete relief should receive a fee award equal to the lodestar amount, while a plaintiff who prevailed on all of his claims but won only limited or partial relief might appropriately receive less than the full lodestar amount. In either case, “the most critical factor [in setting an appropriate fee] is the degree of success obtained” in the litigation as a whole. Id. at 436, 103 S.Ct. at 1941.
Claims can be sufficiently interrelated to require joint consideration for Hensley purposes but sufficiently discrete to be severa-ble from one another on the merits under Rule 54(b) for purposes of appellate jurisdiction. The “common core of facts” or “based on related legal theories” test set out in Hensley suggests a very broad standard of relatedness similar to the one generally used for purposes of claim preclusion.1 The Supreme Court, however, has held that some transactionally related claims are sufficiently discrete to be sever-able from one another under Rule 54(b). See Cold Metal Process Co. v. United Engineering & Foundry Co., 351 U.S. 445, 452, 76 S.Ct. 904, 909, 100 L.Ed. 1311 (1956) (“If the District Court certifies a final order on a claim which arises out of the same transaction and occurrence as pending claims, and the Court of Appeals is satisfied that there has been no abuse of discretion, the order is appealable.”); Sears, Roebuck & Co. v. Mackey, 351 U.S. 427, 436-37 *133& n. 9, 76 S.Ct. 895, 900-01 & n. 9, 100 L.Ed. 1297 (1956) (upholding the certification of a claim that involved “some” different operative facts from those involved in claims that remained pending). I conclude, therefore, that some claims that “involve a common core of facts or [are] based on related legal theories,” Hensley, 461 U.S. at 435, 103 S.Ct. at 1940, can be severed from one another under Rule 54(b) nonetheless.
Thus, a situation can arise in which a plaintiff has prevailed on one claim, another transactionally related claim remains pending, and the district court has certified the adjudicated claim as final under Rule 54(b).2 At that point in the litigation, it seems to me that the analysis required by Hensley cannot be done properly. Because the adjudicated claim is interrelated with the pending one, the hours reasonably spent litigating both must be included in the lodestar. However, because the plaintiffs overall degree of success is as yet unknown (because of the uncertain outcome of the related claim that remains pending), the district court cannot make the requisite comparison between the hours expended and the overall relief obtained. In short, an unmodified Hensley analysis is impossible.
Hensley, however, can easily be modified to make the requisite calculation possible. At the time of the Rule 54(b) certification, the Hensley analysis can be performed on the assumption that the plaintiff will obtain no relief on the claim that still remains pending. Then, the “overall relief obtained” for purposes of the fee determination is simply the relief obtained from having prevailed on the certified claim. That relief can be compared against the total number of hours spent litigating both claims until the time of the certification, and an appropriate fee set accordingly.
If the pending claim also provides for statutory attorneys’ fees and the plaintiff later prevails on it as well, another calculation under Hensley would then be warranted. The effect of the Rule 54(b) certification, I believe, is to sever the previously related claims, both for appellate jurisdiction and for Hensley purposes. Before the certification, “[m]uch of counsel’s time [was] devoted generally to the litigation as a whole, making it difficult to divide the hours expended on a claim-by-claim basis.” Hensley, 461 U.S. at 435, 103 S.Ct. at 1940. After the certification, however, counsel’s time is easily allocable between the two claims: obviously, time spent litigating in the district court is attributable only to the claim still pending. Thus, a Hensley calculation can be made at this second stage by comparing all potentially compensable hours — which include all time spent litigating the pending claim after the certification and all time spent litigating before the certification (which is fairly attributable to both claims) that went uncompensated under the first Hensley calculation — against the degree of additional success afforded by the later judgment (or settlement).
II.
The exact mechanics of all the various permutations of this bifurcated Hensley analysis are not before us at this time. Nonetheless, the result we endorse today makes sense only to the extent that the bifurcated Hensley analysis that it will necessitate is feasible. Feasible it is, although even this brief outline of the analysis reveals its principal vices: (1) it further complicates an already overly complicated field of law, measured against the alternative of requiring a unitary fee determination to be made only after all related claims have been fully adjudicated; (2) it requires some interstitial modification of the procedures set out in Hensley, a binding Supreme Court precedent; and (3) it allows piecemeal appeals of related merits judgments to be followed by piecemeal appeals of related counsel fee awards.
For these reasons, I am uncomfortable with the statement that there is “no policy consideration that would suggest that a *134separate fee award on [the certified merits] judgment should not also be final.” Ante at 131. Nonetheless, I join in the court’s opinion. I do so, despite these misgivings, because I believe that the adverse consequences of postponing finality are sufficiently weighty to overcome the complications of bifurcating the counsel fee determination.
First of all, requiring a unitary fee determination only at the end of the litigation of all related claims would make the Hensley analysis simpler as a formal matter, but often more difficult practically. Frequently claims are litigated over a number of years, and a pending claim can drag on for several years after a related claim has been decided and certified. Memories (unlike pending claims) fade with time, so a district court is surely in a better position to set an appropriate fee relatively contemporaneously, even after accounting for the technical complications engendered by the bifurcation of the Hensley analysis.
Second, a prime policy underlying the various statutory fee provisions is to encourage competent attorneys to file classes of suits that Congress has deemed to be in the public interest. See, e.g., City of Riverside v. Rivera, 477 U.S. 561, 575-81, 106 S.Ct. 2686, 2692-98, 91 L.Ed.2d 466 (1986) (plurality opinion) (discussing the rationale behind 42 U.S.C. § 1988). To the extent that the calculation of a counsel fee award is postponed, the attorney must wait that much longer to receive his compensation. Money has a time value, and attorneys often have cash flow considerations to worry about. For these reasons, an award delayed is an award reduced — and with it, the incentive to bring similar suits in the future. At the margin, therefore, delaying the fee calculation will result in fewer actions, thus undermining the purpose underlying the relevant statutory fee provision.
Although for me the question is close, I conclude that on balance the benefits of the rule we announce today outweigh its drawbacks. I understand nothing in the court’s analysis to be inconsistent with the views I have expressed and, on that assumption, I join in the court’s opinion.
. Compare, e.g., Fed.R.Civ.P. 13(a) (counterclaim must be raised "if it arises out of the transaction or occurrence that is the subject matter of the opposing party’s claim”); Restatement (Second) of Judgments § 24 & comments a, b (1982) (equating "claim” with “transaction” and defining "transaction” in terms of “a natural grouping or common nucleus of operative facts”).
. Indeed such a situation arose earlier in this litigation: the plaintiffs prevailed on their APA claim; the § 1983 claim, which arose out of a common core of operative facts as the APA claim, remained pending; and the district court certified the APA claim as final nonetheless.