Appellants are 25 law firms (“Class Counsel”) who represented bondholders (“Class Plaintiffs”) in the largest municipal bond default in history. From 1977 to 1983, the Washington Public Power Supply System (‘WPPSS”) sold bonds with a face value of $2.25 billion to finance construction of two nuclear power plants. The plants were never completed and WPPSS defaulted on its bond payments. In 1983, purchasers of the bonds filed a class action against WPPSS and nearly 200 other defendants alleging violations of state and federal securities laws in the sale of the bonds.1 The claims of class members totalled almost $1.47 billion. The ease was ultimately resolved through 22 separate settlement agreements, which created a settlement fund of $687 million.
Class Counsel requested attorneys’ fees totalling $103 million from the settlement fund, which they asserted was a reasonable fee under either the percentage-of-the-fund method or the lodestar/multiplier approach.2 See In re Washington Pub. Pow*1295er Supply Sys. Sec. Litig., 779 F.Supp. 1063, 1084 (D.Ariz.1990) [hereinafter WPPSS II]. Under the lodestar approach, Class Counsel reached their $103 million figure by enhancing a lodestar of nearly $33 million by various multipliers which, in the aggregate, constituted a “blended” multiplier of 3.1. In the alternative, Class Counsel claimed that their $103 million lodestar figure would be reasonable under the percentage method because it represented only 13.6 percent of the settlement fund. Id. at 1081.
Employing the lodestar rather than the percentage method, the district court made certain reductions from Class Counsel’s $103 million figure, ultimately arriving at a lodestar of $27 million. The court denied most of Class Counsel’s requests for multipliers, but enhanced the awards to eleven individual attorneys for the exceptional quality of their representation. These individual enhancements resulted in a blended multiplier of 1.2, which yielded a final award of $32 million.
Class Counsel moved for reconsideration of the fee award, stating that their “principal assertion ... is that larger multipliers should be awarded.” Reply Memorandum of Class Counsel in Support of Reconsideration on Attorneys’ Fee Order at 7. They also explained that they were arguing “for a percentage as an alternative method, as a way to test the Kerr [lodestar/multiplier] analysis and to confirm the reasonableness of the proposed multipliers.” Id. After reconsidering Class Counsel’s arguments, the district court refused to increase its initial fee award of $32 million. In re Washington Pub. Power Supply Sys. Sec. Litig., 779 F.Supp. 1056, 1063 (D.Ariz.1991) [hereinafter WPPSS III ]. Class Counsel appeal this award as unreasonably low. Their principal argument on appeal is that the district court neglected to enhance the lodestar for two of the Kerr factors — risk of nonpayment and results obtained — when calculating their fee. See Kerr v. Screen Extras Guild, Inc., 526 F.2d 67, 70 (9th Cir.1975).
I
Is the Percentage Method Required in Common Fund Cases?
At the outset, Class Counsel urge us to follow the Eleventh Circuit’s lead in mandating the use of the percentage method in common fund cases. See Camden I Condominium Ass’n v. Dunkle, 946 F.2d 768, 774 (11th Cir.1991).3 Because the law in our circuit is settled on this issue, we are not at liberty to follow the Eleventh Circuit. We instead apply the law of our circuit that the district court has discretion to use either method in common fund cases.
In Florida v. Dunne, where we approved the district court’s use of the lodestar method to award fees in a common fund case, we explained:
Despite the recent ground swell of support for mandating a percentage-of-the-fund approach in common fund cases, however, we require only that fee awards in common fund cases be reasonable under the circumstances. Accordingly, either the lodestar or the percentage-of-the-fund approach “may, depending upon the circumstances, have its place in determining what would be reasonable compensation for creating a common fund. ”
915 F.2d 542, 545 (9th Cir.1990) (quoting Paul, Johnson, Alston & Hunt v. Graulty, 886 F.2d 268, 272 (9th Cir.1989) (emphasis added)). In so holding, Dunne relied on two previous cases in which we approved use of the percentage method in a common fund context, but made it clear that either the percentage or the lodestar method may be appropriate depending on the circumstances. See Six Mexican Workers v. Arizona Citrus *1296Growers, 904 F.2d 1301, 1311 (9th Cir.1990) (“the choice between lodestar and percentage calculation depends on the circumstances”); Graulty, 886 F.2d at 272 (same).
Although Class Counsel ultimately acknowledge, as they must, that we have “not yet adopted Camden’s rale that the percentage method is mandatory,” they make the further argument that Ninth Circuit law mandates use of the percentage method in common fund cases unless the district court finds “special considerations” that warrant use of the lodestar method. Appellants’ Opening Br. at 70. Thus, they contend that the district court erred as a matter of law in choosing the lodestar method without first identifying “any 'special circumstances’ that would overcome the presumption in favor of the percentage method.” Appellants’ Reply Br. at 37.
As authority for their argument, Class Counsel seize upon our statement in Six Mexican Workers that “[although statutory awards of attorneys’ fees are subject to ‘lodestar’ calculation procedures, a reasonable fee under the common- fund doctrine is calculated as a percentage of the recovery.” 904. F.2d at 1311 (citing Blum v. Stenson, 465 U.S. 886, 900 n. 16, 104 S.Ct. 1541, 1550 n. 16, 79 L.Ed.2d 891 (1984)). We went on to say that a percentage award “should be adjusted, or replaced ... when special circumstances indicate that the percentage recovery would be either too small or too large in light of the hours devoted to the case or other relevant factors.” Id. Class Counsel read too much into these two passages from Six Mexican Workers. Whatever uncertainty in our law these passages may have created, it was resolved in the remainder of the paragraph when we reaffirmed our settled rule that either the lodestar or percentage method may “‘have its place in determining what would be reasonable compensation for creating a common fund,’ ” id. (quoting Graulty, 886 F.2d at 272), and that “the choice between lodestar and percentage calculation depends on the circumstances.” Id. This clarification of our law was sharpened by Dunne, where we approved the district court’s use of the lodestar method even though the case involved no special circumstances that might have made the percentage method inappropriate. 915 F.2d at 545. Given this clear circuit precedent, we reject appellants’ argument that the district court’s refusal to use the percentage method constitutes an error of law. This was not a ease, like Graulty, where “it was impractical, if not impossible, to use the lodestar approach.”4 Id. at 545-46. Here, no such complexity exists to make the calculation of the lodestar impractical.5 Accordingly, we restate the law of our circuit that, in common fund cases, no presumption in favor of either the percentage or the lodestar method encumbers the district court’s discretion to choose one or the other. As always, when determining attorneys’ fees, the district court should be guided by the fundamental principle that fee awards out of common funds be “reasonable under the circumstances.” Id. at 545 (emphasis added).
II
Did the District Court Abuse Its Discretion in Applying the Lodestar Method Instead of the Percentage Method?
Class Counsel contend that even if the district court had the discretion to choose between the lodestar and percentage methods, it abused that discretion by choosing the lodestar method instead of the percentage method in this particular case. We review the district court’s award of attorneys’ fees, including its choice of methods, for abuse of discretion. Drucker v. O’Brien’s Moving & *1297Storage, Inc., 963 F.2d 1171, 1172 (9th Cir.1992).
Class Counsel’s basic contention is that the district court erred, in considering the size of the fund ($687 million) when deciding which method was more appropriate in this case. They argue that the district court should have awarded them 13.6 percent of the fund, regardless of its size, because that figure was lower than both our circuit’s 25 percent “benchmark” figure, see Graulty, 886 F.2d at 272, and the 20-30 percent awards “typical” in common fund cases. Appellants’ Opening Br. at 9-10, 57, 73. We are unpérsuaded.
In explaining its decision to use the lodestar method, the district court first observed that “the immense size of the settlement fund in this action far exceeds the amount of settlement funds in virtually all similar cases” reviewed by the court. WPPSS II, 779 F.Supp. at 1085. Because of this, the court concluded that the size of the fund magnified beyond all reasonable limits the margin of error inherent in a percentage fee award. Id. at 1084-85. The court acknowledged Class Counsel’s central proposition that a fee in the range of 20-40 percent is typical in many common fund cases, but concluded that the size of the fund here precluded a meaningful analysis of a percentage fee award in terms of comparisons with awards in other reported cases. Id. at 1085. The court’s review of relevant cases showed that in many eases awards fall outside the “typical” range and that the percentage of an award generally decreases as the amount of the fund increases. Id. In light of this, the court stated that “[g]iven the variances in the sizes of settlement funds, Class counsels’ data, intended to demonstrate that the percentage they request is far less than the norm, aptly demonstrate the virtually impossible task of setting any particular percentage as a 'proper one.” Id. (emphasis in original). Thus, the court found no rational basis for Class Counsel’s request for 13.6 percent of the fund, concluding that the figure was “arbitrary” because Class Counsel could just as easily have requested 3.6 percent or 36.1 percent. Id.; see also In re Superior Beverage/Glass Container Consol. Pretrial, 133 F.R.D. 119, 125 (N.D.Ill.1990) (“There is no necessary logical connection between percentages and reasonable compensation. At best, percentages are simply a rough and ready way of estimating contingency plus effort.”). Even after reconsidering Class Counsel’s arguments, the district court refused to “retreat from its lodestar analysis,” which it believed would provide the fairest, most reasonable fee award. WPPSS III, 779 F.Supp. at 1060.
We agree with the district court that there is no necessary correlation between any particular percentage and a reasonable fee. With a fund this large, picking a percentage without reference to all the circumstances of the case, including the size of the fund, would be like picking a number out of the air. As the district court, in the context of rejecting a results multiplier, explained:
[i]n determining fee awards in class actions, it is especially important that judges not be unduly influenced by the monetary size of the settlement. A sizable settlement can reflect a number of factors in addition to the prestige, skill and vigor of Class counsel. Thus, it is imperative that the amount of th[e] settlement ... be viewed from the proper perspective and in the context of all relevant circumstances.
WPPSS II, 779 F.Supp. at 1097-98 (citations omitted). Because a court must consider the fund’s size in light of the circumstances of the particular case, we agree with the district court that the 25 percent “benchmark” is of little assistance in a case such as this.
In fact, the reasonableness of the district court’s choice of the lodestar approach is attested to by the methodology used by Class Counsel in their own fee petition. Class Counsel requested a fee of $103 million based upon a lodestar enhanced by a 3.1 blended multiplier. They then defended the reasonableness of the $103 million request by pointing out that it represented only 13.6 percent of the settlement fund- of $687 million. In other words, they did not base their fee request on a percentage-of-the-fund calculation, but argued that their enhanced lodestar fee of $103 million was necessarily reasonable because it represented only half the fee that would be awarded if the “benchmark” of 25 *1298percent were applied.6 Not only does this type of circular reasoning point up the danger of applying percentages indiscriminately in awarding fees, it also highlights the disparity between Class Counsel’s fee petition, where they relied primarily on the lodestar approach, and their appellate brief, where they contend the district court abused its discretion in using the lodestar rather than the percentage approach.
It is not difficult to demonstrate why courts cannot rationally apply any particular percentage — whether 13.6 percent, 25 percent or any other number — in the abstract, without reference to all the circumstances of the case. To illustrate the point, we need only assume that the WPPSS bond issue was $4.5 billion instead of $2.25 billion, and that the settlement fund was $1.4 billion rather than roughly $700 million. Assume as well that all other variables remained constant— the merits of plaintiffs’ case on the facts and the law, the skill and time of counsel required to develop the merits of the case in the litigation process, and counsel’s hourly rates. Would Class Counsel still contend that 13.6 percent was a reasonable figure? An award of 13.6 percent of the fund would give Class Counsel a fee of $200 million, double the fee they actually seek for their effort in this case. Plainly, a fee of $200 million for the same effort by counsel with the same level of skill would be a windfall rather than a reasonable fee. In sum, the district court was correct that there is nothing inherently reasonable about an award of 13.6 percent of a fund regardless of its size. Accordingly, we hold that the district court acted well within the bounds of its discretion in considering the size of the fund in choosing where the lodestar rather than the percentage approach to arrive at a reasonable fee.
Ill
Did the District Court Err in Reducing the Lodestar Hours ?
Class Counsel make various arguments challenging the district court’s reduction of their submitted billable hours from 157,000 to 137,000. First, they argue that the district court’s reductions in their lodestar hours for a variety of reasons, including duplicative work and vague time sheet entries, were arbitrary. They are particularly critical of the reductions in the hours of Messrs. Schulman, Berman, and Klafter, who were “key lieutenants” of lead counsel who “lived and breathed this case,” often working 14-16 hour days. Appellants’ Opening Br. at 96-97. After carefully scrutinizing Class Counsel’s time records, the district court cut the lodestar hours to reflect its finding that some work was duplicative and that excessive time was spent on certain activities. As for Messrs. Schulman, Berman, and Klafter in particular, the court provided specific examples to justify its 12-15 percent cuts in their hours. See WPPSS II, 779 F.Supp. at 1105-06 (Klafter), 1106 (Berman), 1114-16 (Schul-man). We hold that the district court acted within the bounds of its discretion in reducing the lodestar for unnecessary and duplica-tive work.
Second, Class Counsel challenge the reductions in Mr. Weiss’ teleconference hours as arbitrary. However, the court reduced Mr. Weiss’ teleconference hours from 1000 to 900 because he did not adequately document the substance of his calls and repeatedly rounded his time upward. Id. at 1113-14. These reductions were not an abuse of discretion.
Third, Class Counsel contend that the district court unjustifiably halved the travel time of all attorneys. They argue that “[t]he court’s presumption that these activities were not necessary or beneficial to the Class is grossly unfair” and finds no support in the record. Appellants’ Opening Br. at 97. In addition, they stress that extensive travel, especially by lead counsel, is an inevitable consequence of this type of lawsuit. Id. at 98. The district court reduced attorney travel time after finding that the attorneys gen*1299erally billed the entire duration of time spent in transit. See In re “Agent Orange” Prod. Liab. Litig., 818 F.2d 226, 237-38 (2d Cir.1987) (finding appropriate across the board percentage cuts to “trim[] fat from a fee application” where fee petitions were voluminous). The court reasoned that the distractions associated with travel, especially after a full day of work, likely reduced the attorneys’ effectiveness while en route. WPPSS II, 779 F.Supp. at 1094-95. We hold that the court did not abuse its discretion in reducing attorney travel time for the reasons given.
Finally, Class Counsel argue that they should be compensated for time spent attempting to obtain a reasonable fee because “the interests of this and future classes will be served if counsel are compensated adequately for their services.” Appellants’ Reply Br. at 44 n. 49. The district court held that charges for time spent preparing fee petitions “are inappropriate in common fund cases” because such an undertaking “eon-fer[s] no benefit on the Class.” WPPSS II, 779 F.Supp. at 1096. We agree. Time spent obtaining an attorneys’ fee in common fund cases is not compensable because it does not benefit the plaintiff class. See City of Detroit v. Grinnell Corp., 560 F.2d 1093, 1102 (2d Cir.1977); Lindy Bros. Builders, Inc. v. American Radiator & Standard Sanitary Corp., 540 F.2d 102, 110-11 (3d Cir.1976) (en banc); In re Nucorp Energy, Inc., 764 F.2d 655, 661 (9th Cir.1985) (citing Grinnell with approval in dicta).
IV
Did the District Court Have Discretion to Apply a Risk Multiplier in a ■Common Fund Case?
Next we address Class Counsel’s argument that the district court abused its discretion in refusing to enhance the lodestar for taking the risk of losing and coming up empty-handed for their services. Class Plaintiffs-Appellees counter that City of Burlington v. Dague, — U.S. -, 112 S.Ct. 2638, 120 L.Ed.2d 449 (1992), which proscribed contingency enhancements in statutory fee-shifting cases, should be extended to proscribe contingency enhancements in common fund cases.7 We agree with Class Counsel, however, that Dague’s rationale for barring risk multipliers in statutory fee cases does not operate to bar risk multipliers in common fund cases.8
It is an established practice in the private legal market to reward attorneys for taking the risk of non-payment by paying them a premium over their normal hourly rates for winning contingency cases. See Richard Posner, Economic Analysis of Law § 21.9, at 534-35 (3d ed. 1986). Contingent fees that may far exceed the market value of the services if rendered 'on a non-contingent basis are accepted in the legal profession as a legitimate way of assuring competent representation for plaintiffs who could not afford to pay on an hourly basis regardless whether they win or lose. See Model Rules ofProfes-*1300sional Conduct Rule 1.5(a)(8) (1992); Model Code of Professional Responsibility DR 2-106(B)(8) (1980); Canons of Ethics § 12, 33 A.B.A.Rep. 575, 578 (1908). As the court observed in Behrens v. Wometco Enter., Inc., 118 F.R.D. 534, 548 (S.D.Fla.1988), aff'd, 899 F.2d 21 (11th Cir.1990), “[i]f this ‘bonus’ methodology did not exist, very few lawyers could take on the representation of a class client given the investment of substantial time, effort, and money, especially in light of the risks of recovering nothing.” And in In re Union Carbide Corp. Consumer Prods. Business Sec. Litig., 724 F.Supp. 160, 169 (S.D.N.Y.1989), the court addressed the negative impact the proscription of risk multipliers might have on the important goal of furthering the purposes of federal securities laws:
[I]ndividuals damaged by violations of the federal securities laws should have reasonable access to counsel with the ability and experience necessary to analyze and litigate complex eases.... A large segment of the public might be denied a remedy for violations of the securities laws if contingent fees awarded by the courts did not fairly compensate counsel for the services provided and the risks undertaken.
For these reasons, courts have routinely enhanced the lodestar to reflect the risk of nonpayment in common fund cases.9
In Dague, the Court held that the private market’s model of paying premiums for the risk of non-payment was inapplicable in statutory fee-shifting cases. For a variety of reasons, the Court concluded that the private market practice of rewarding attorneys for taking cases on a contingency basis would unduly burden losing parties who are required by statute to pay no more than a “reasonable” fee for the services rendered by the winning party’s attorneys in that particular case. Dague, — U.S. at -, 112 S.Ct. at 2641. For example, the Court reasoned that it would be incompatible with fee-shifting statutes to burden losing parties with a contingency enhancement that has the effect of compensating the winning attorneys for time gambled away in the contingency eases they lose:
An attorney operating on a contingency-fee basis pools the risks presented by his various cases: cases that turn out to be successful pay for the time he gambled on those that did not. To award a contingency enhancement under a fee-shifting statute would in effect pay for the attorney’s time (or anticipated time) in cases where his client does not prevail.
Id. at-, 112 S.Ct. at 2643 (emphasis in original). This concern, along with concern over the complexity and cost of administering a system of contingency enhancements in the statutory fee context, drove the Court to the conclusion that “[i]t is neither necessary nor even possible for application of the fee-shifting statutes to mimic the intricacies of the fee-paying market in every respect.” Id.
As we read Dague, the concerns that drove the Court to reject contingency enhancements in the statutory fee context apply with much less force in common fund cases. Unlike statutory fee-shifting cases, where the winner’s attorneys’ fees are paid by the losing party, attorneys’ fees in common fund cases are not paid by the losing defendant, but by members of the plaintiff class, who shoulder the burden of paying their own counsel out of the common fund. See Boeing Co. v. Van Gemert, 444 U.S. 472, 478-79, 100 S.Ct. 745, 749, 62 L.Ed.2d 676 (1980); Vincent v. Hughes Air West, Inc., 557 F.2d 759, 769-70 (9th Cir.1977). There is nothing unfair about contingency enhancements in common fund cases because of the equitable notion that those who benefit from the creation of the fund should share the wealth with the lawyers whose skill and effort helped create it. Boeing, 444 U.S. at 478, 100 S.Ct. at 749; Graulty, 886 F.2d at 271. As the Seventh Circuit observed in Skelton v. General Motors Corp., 860 F.2d 250, 254 (7th Cir.1988), “in the common fund context, attorneys whose compensation depends on their winning the case, must make up in compensation *1301in the cases they win for the lack of compensation in the cases they lose.”10
Thus, the concerns expressed in Pague about unduly burdening losing parties in statutory fee cases are not present in common fund cases where .fees are paid out of the settlement fund. How the fund is divided between, members of the class and class counsel is of no concern whatsoever to the defendants who contributed to the fund. “In a common fund case, where 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.” Skelton, 860 F.2d at 254. Accordingly, because we find Dague’s reasoning inapposite in the common fund context, we hold that district courts have discretion to use risk multipliers to enhance the lodestar in common fund cases.
V
Did the District Court Abuse Its Discretion in Refusing to Award a Risk Multiplier in This Case?
Having determined that the district court had discretion to award a risk multiplier-in this common fund case, we must now decide whether, as Class Counsel contend, the court abused its discretion in refusing to award a risk multiplier in this ease.11 We hold that it did, and remand for reconsideration of the risk multiplier issue.
The district court twice determined that it had discretion to award a risk multiplier, but each time it refused to do so. In its initial fee decision, the court stated that its discretion to award a multiplier was “restricted” to only those cases in which the plaintiffs faced “substantial difficulties in finding counsel in the local or other relevant market without an enhancement for risk.” WPPSS II, 7.79 F.Supp. at 1090. The court declined to award a risk multiplier after finding that Class Plaintiffs had little difficulty obtaining counsel. Id. at 1090, 1101.
On reconsideration, the district court again stated that it “had at its disposal the necessary discretion to apply multipliers,” although it disapproved its earlier language that its discretion was “restricted.” WPPSS III, 779 F.Supp. at 1061. The district court again denied a risk multiplier, however, reasoning that law firms showed no reluctance to take the case and that the firms’- high hourly rates already reflected some of the risk in the case:
Although the risk was high, it was not so high that firms were reluctant to take the case. Indeed, the opposite is true. The Court is not persuaded that a multiplier expectancy acted as a great incentive. That claim rings hollow in light of the high hourly fees customarily charged in these cases. Surely, risk is reflected, at least in part, by the hourly fees charged.
Id. (footnote omitted).
We find no basis in the record to support the district court’s finding that Class Counsel’s decision to represent the class was not driven, at least in part, by an expectancy that their fee would be enhanced if they were successful. We also find no basis in the record for the court’s statement that the fees charged by Class Counsel reflected, in part, the risk of non-payment. In fact, uncontro-verted affidavits submitted by Class Counsel are to the contrary.
First, the affidavits show that many firms sought to represent Class Plaintiffs based on an expectation that a substantial multiplier (or percentage fee) would be awarded for a *1302favorable result. See, e.g., Irwin Aff. ¶ 6; Pym Decl. ¶¶ 5-6; Meehan Aff. ¶ 5; Molloy Aff. ¶¶ 3-4; Weiss Decl. ¶¶ 5, 8, 14; see also Cotchett Decl. ¶¶ 6-10; Byrnes Aff. ¶¶ 5-7; Faucher Decl. ¶ 5; Reasoner Decl. ¶ 5; Burke Aff. ¶ 2. Even though some of these firms customarily do little contingent fee work, they took the risk of representing Class Plaintiffs in this case because of an expectation that a favorable result would produce an enhanced fee. See Irwin Aff. ¶¶ 3-6; Pym Decl. ¶¶ 3-6; Meehan Aff. ¶¶ 5, 7. Two firms turned down opportunities to represent defendants in this case because of their expectation that, although representing Class Plaintiffs had its risks, it also offered the potential of a substantially enhanced fee. Without the expectation of enhancement for risk, the firms would have chosen to represent defendants on an hourly basis. See Pym Decl. ¶¶3-6; Meehan Aff. ¶¶4-7. These expectations appear to have been reasonable, especially given the common practice in the 1980s for courts to award multipliers in common fund cases. See, e.g., In re Activision Sec. Litig., 723 F.Supp. 1373, 1377-78 (N.D.Cal.1989) (documenting common fund cases where fees were enhanced through use of multiplier or percentage method); Behrens, 118 F.R.D. at 549 (same).
Second, Class Counsel’s affidavits show that the hourly rates charged by many of the firms did not reflect the risk of non-payment. For example, Mr. Weiss stated that the rates billed in his firm’s lodestar submission were the same as its “market rates when performing its legal services on a risk-free, non-contingent basis.” Weiss Decl. ¶ 2. Similarly, Mr. Meehan declared that his firm “utilized billing rates ... which were established for purposes of charging fees to clients that pay on a monthly, non-contingent basis. The payment of the lodestar at these hourly rates represents no more than what my firm would have received” had it accepted an offer to represent a defendant on a non-contingent basis. Meehan Aff. ¶ 6; see also Pym Decl. ¶ 6.12
The district court several times acknowledged the riskiness of this case. See WPPSS I, 720 F.Supp. at 1391 (the case was “fraught with risk and recovery was far from certain”); WPPSS II, 779 F.Supp. at 1091 (noting that Class Counsel prosecuted this “rare and exceptional” case in “the face of uncertain victory” and “with absolutely no guaranty of payment”); WPPSS III, 779 F.Supp. at 1061 (stating that the risk was “high”). We find no evidence in the record as it now exists that supports the district court’s reasoning in denying a risk multiplier. Accordingly, we hold that the district court abused its discretion in denying a risk multiplier. We vacate the fee award and remand for further consideration of the issue.
We recognize that as the record now stands, Class Counsel’s affidavits are uncontroverted. The explanation may be that Class Plaintiffs were not represented by counsel in the fee-setting proceedings before the district court. Because in common fund cases the relationship between plaintiffs and their attorneys turns adversarial at the fee-setting stage, courts have stressed that when awarding attorneys’ fees from a common fund, the district court must assume the role of fiduciary for the class plaintiffs. See Skelton, 860 F.2d at 253; Grinnell, 560 F.2d at 1099. As the district court here correctly observed, at the fee-setting stage, “[p]lain-tiffs’ counsel, otherwise a fiduciary for the class, has become a claimant against the fund created for the benefit of the class. It is obligatory, therefore, for the trial court judge to act with ‘a jealous regard to the rights of those who are interested in the fund’ in determining what a proper fee award is.” WPPSS II, 779 F.Supp. at 1083 (citing Trustees v. Greenough, 105 U.S. 527, 536, 26 L.Ed. 1157 (1882)). Thus, on remand, the district court here need not be limited to the existing evidentiary record on the risk multiplier issue; it has the discretion to take steps to develop the record more fully before deciding the issue, including the discretion to appoint counsel to represent Class Plaintiffs for this limited purpose.
VI
Did the District Court Abuse Its Discretion in Denying a Results Multiplier?
Class Counsel argue that the district court abused its discretion in rejecting their *1303request for a lodestar enhancement for the “extraordinary” and “excellent” results they achieved in this case. Appellants’ Opening Br. at 90-91. Their claim that the results were “extraordinary” and “excellent” is based primarily on the size of the settlement fund. Appellants’ Reply Br. at 28-32; Appellants’ Opening Br. at 91-93. They stress that the fund of $687 million represents the largest recovery ever in a securities class action. Appellants’ Reply Br. at 30. Moreover, they contend that “a recovery of at least 40 cents on the dollar in a difficult and risky securities fraud case such as this is superlative,” and warrants a substantial enhancement for result. Id. at 31 (footnote omitted). Thus, Class Counsel contend that the district court erred by refusing to enhance the award for results as part of their requested blended multiplier of 3.1, or by simply awarding them 13.6 percent of the fund (which equals their enhanced lodestar of $103 million). Once again, Class Counsel stress that both their blended multiplier and percentage figure fall within, or below, the normal ranges in other common fund cases.
In rejecting the request for a results multiplier, the district court refused to consider the size of the fund in isolation. The court heeded the oft-repeated warning in Grinnell that “[i]n determining awards in class actions, it is especially important that judges not be unduly influenced by the monetary size of the settlement” and stated that it would view the size of the fund “in the context of all relevant circumstances.” WPPSS II, 779 F.Supp. at 1097-98 (citing Grinnell, 560 F.2d at 1099).
First, the district court acknowledged that the settlement amount was “enormous.” Id. at 1098. The court explained, however, that the money actually available for distribution to Class Plaintiffs was somewhat less than the settlement fund’s stated value of $687 million. After factoring in deductions for certain non-class parties, including the indenture trustee, Chemical Bank, the court said that “a fairer statement may be that ... the amount available to Class members and to counsel for their fees and expenses [] is approximately $590 million plus some interest.” Id. (footnote omitted).
Even more important to the district court was the reality that, while the fund amount was extremely high, so too were the losses suffered by Class Plaintiffs — almost $1.47 billion on principal exclusive of lost interest. “Because the bonds were purchased for the interest they promised, this [figure] gravely understates the monetary devastation that many Class members feel.” Id. Thus, despite the undeniably large' size of the fund, the district court was “faced with the inescapable fact that the extraordinary amount appears far more modest when viewed in relation to the injuries sustained by Class members, particularly those many individuals who invested their life savings.” Id.
On reconsideration, the district court elaborated on its reasoning by pointing out that the quality of the Class Counsel created an expectation of “excellent results.” WPPSS III, 779 F.Supp. at 10.61. The court described Class Counsel as representing “the highest echelon in the securities’ litigation bar,” and that “[e]ommensurate with their skill, and reputation comes an expectancy of results that exceed average levels. Counsel demand and receive the highest fees because they obtain extraordinary results. Those very fees formed the basis of this Court’s lodestar calculation_” Id. at 1062. In considering the totality of the circumstances, the district court concluded that the results in this ease did not “exceed the extraordinary.” Id.
We agree with the district court that the large size of the settlement fund obtained by Class Counsel relative to recoveries in other cases does not, ipso facto, mean that it is an extraordinary result. Accordingly, we hold that the district court did not exceed the bounds of its discretion in denying a results multiplier after concluding that the results in this case, when viewed in the context of all relevant circumstances, did not “exceed the extraordinary.” Id.
While Class Counsel argue that the result obtained is “extraordinary” because of its size, they also appear to quarrel with the district court’s use of the “exceptional success” standard articulated in the statutory fee case Blum v. Stenson, 465 U.S. 886, 900-01, 104 S.Ct. 1541, 1549-50, 79 L.Ed.2d 891 *1304(1984) (holding that in a statutory fee case the results obtained factor was already “subsumed” in the lodestar and that results multipliers should be awarded only in some “cases of exceptional success”); see also Stewart v. Gates, 987 F.2d 1450, 1458 (9th Cir.1993) (statutory fee case holding similarly). Their argument vaguely asserts that the district court erred by relying on a handful of statutory fee eases to support its use of this demanding standard. Appellants’ Opening Br. at 90. Once again, as they did with the district court’s refusal to award a risk multiplier, Class Counsel view the district court’s application of legal principles developed in the statutory fee context to a common fund case as its “central legal error.” Id. at 59, 90-91. Yet they offer no cogent argument why courts should not selectively apply legal principles developed in statutory fee cases if those principles also happen to be logically applicable to the task of arriving at a reasonable fee in a common fund case. They also fail to provide an alternative standard for the court to use in common fund cases other than that “[a]n excellent result should support a larger fee.” Id. at 91. That very well may be true, but only so long as the fee would be reasonable under the circumstances of the particular case.
We fail to grasp the logic of Class Counsel’s mantra-like argument that principles applicable in statutory fee cases are necessarily inapplicable in common fund cases. Whether a legal principle used in one context (a statutory fee ease) is logically applicable in a different context (a common fund case) depends on whether the principle makes sense in the new context. It would defy common sense to adopt a per se rule, as Class Counsel would have us do, that principles' developed in statutory fee cases are, ipso facto, inapplicable in common fund cases. This ease is illustrative. For example, we agree with Class Counsel that Da-gue’s proscription of risk multipliers in statutory fee cases should not be extended to common fund cases. See Part IV supra. However, we disagree with Class Counsel’s contention that the district court erred by using in this common fund case the “exceptional success” standard developed for results multipliers in statutory fee cases. The district court itself acknowledged that Blum and other cases which used the “exceptional success” standard were statutory fee cases. WPPSS II, 779 F.Supp. at 1089. Nevertheless, the court found the decisions helpful “insofar as they address the reasonableness of a fee award” in the common fund context. Id. Specifically, the court implicitly found that the reasoning in the statutory fee cases supported its own conclusion that Class Counsel’s hourly rates, used to calculate the lodestar, already reflected an expectancy of “excellent” or “extraordinary” results. WPPSS III, 779 F.Supp. at 1061-62; see pages 1303-1304 supra. Accordingly, we hold that the district court’s application of the “exceptional success” standard in this case was not an abuse of discretion.
VII
Did the District Court Abuse Its Discretion in Awarding Quality of Representation Multipliers to Individual Attorneys Rather Than to Their Firms?
Class Counsel argue that the district court erred by awarding quality of representation multipliers to 11 of the 291 attorneys who represented Class Plaintiffs for their exceptional efforts.13 Instead, Class Counsel contend that multipliers should be awarded to entire firms, not to individual attorneys, although they offer no authority for this proposition. Appellants’ Opening Br. at 94-95.
The authority that does exist on the issue is contrary to Class Counsel’s position. The practice of awarding quality of representation multipliers to individual attorneys has been approved of in several cases. See, e.g., In re ‘Agent Orange”, 818 F.2d at 234-35; In re Equity Funding Corp. Sec. Litig., 438 F.Supp. 1303, 1337 (C.D.Cal.1977) (stating that the court must “analyze the quality of the performance of the various lawyers individually” and awarding a range of multipliers (from three to one) to individual attorneys with different levels of involvement in the *1305case). We do not find this practice problematic, and hold that the district court did not abuse its discretion in awarding quality of representation multipliers to individual attorneys rather than to firms.
VIII
Did the District Court Abuse Its Discretion by Failing to Compensate Adequately the Gold Firm for Delay in the Payment of Its Fees?
The Gold firm argues that the district court did not adequately compensate it for the seven year delay in payment of its fees. The district court has discretion to compensate delay in payment in one of two ways: (1) by applying the attorneys’ current rates to all hours billed during the course of the litigation; or (2) by using the attorneys’ historical rates and adding a prime rate enhancement. Skelton, 860 F.2d at 255 n. 5. In this ease, the court appeared to use the current rate method, applying current hourly rates to all hours spent on the litigation by attorneys still at the Gold firm when the fee petition was filed. WPPSS II, 779 F.Supp. at 1099-1100, 1145-49. However, the court did not use the current rate method for attorneys who had left the firm prior to the filing of the fee petition. Instead, it applied the last hourly rates those attorneys charged before leaving without adding a prime rate enhancement to compensate for the delay in payment.14 Id.
We agree with the Gold firm that this hybrid method is paradoxical, and hold that its use constituted an abuse of discretion. Accordingly, we vacate the Gold firm’s fee award and remand for recalculation. The district court’s use of current rates for attorneys still at the firm was not improper. But the last rates charged by attorneys who left prior to the fee petition, without a prime rate enhancement, inadequately compensate the firm for the delay in receiving its fees. The time value of money lost by the firm is only partially recouped. Full compensation requires charging current rates for all work done during the litigation, or by using historical rates enhanced by an interest factor. When recalculating the hourly rates for attorneys who left the firm on remand, the district court is, of course, free to use either current rates for attorneys of Comparable ability and experience or historical rates coupled with a prime rate enhancement.
IX
Did the District Court Abuse Its Discretion in Reducing the Berger Firm’s Fees Because Its Time Records Were Deficient?
In support of its fee application, the Berger firm attached computer generated summaries of all hours expended on this litigation. In its initial fee decision, the district court found these records “remarkably deficient,” describing the information contained therein as “indefinite, non-specific, uninformative and generally inadequate.” WPPSS II, 779 F.Supp. at 1140. Accordingly, the court reduced the number of hours submitted by the firm, which resulted in a lodestar for the firm significantly below the requested amount. Id. at 1140-44.
The Berger firm contends that the court abused its discretion in reducing the firm’s lodestar because the court never specified how detailed it expected the supporting documents to be. The Berger firm also argues that the district court erred by not requesting more detailed records after becoming dissatisfied with the computer summaries. Finally, the firm contends that, even if the court did not err initially in reducing the lodestar, it abused its discretion by refusing to reconsider its earlier decision in light of more complete daily time sheets provided by the firm.
The party petitioning for attorneys’ fees “bears the burden of submitting detailed time records justifying the hours claimed to have been expended.” Chalmers v. City of Los Angeles, 796 F.2d 1205, 1210 (9th Cir.1986). Thus, the Berger firm bore the risk of failing to provide adequate back-up documentation for its fee request. As the Supreme Court has said, “[w]here the documentation of hours is inadequate, the district court may reduce the award accordingly.” Hensley v. Eckerhart, 461 U.S. 424, 438, 103 *1306S.Ct. 1933, 1939, 76 L.Ed.2d 40 (1983). It also follows that the district court was neither obligated to explain what type of records should be submitted, nor to request additional information. The burden of presenting the appropriate fee documentation rests squarely on the shoulders of the attorneys seeking the award. Accordingly, we hold that the district court did not abuse its discretion in initially reducing the Berger firm’s lodestar.
However, we believe the court did abuse its discretion in refusing to reconsider its initial decision in light of additional documentation, including five photocopied volumes of handwritten daily time logs, provided by the Berger firm in support of its motion for reconsideration. In denying the motion, the district court said:
The records originally received were computer print outs of daily time records. The newly presented records are the handwritten reports from which those computer records were made. Thus, there is little room for variance. The Court doubts that the new records could be helpful.
WPPSS III, 779 F.Supp. at 1062 (emphasis in original). We believe the court was mistaken in its characterization of the records presented in support of the original petition. The records do not appear to be computer printouts of daily time records, but rather computer printouts of summaries of daily time records. The Berger firm explained this difference to the court in an affidavit attached to its motion for reconsideration, see Berger ER, Tab 4155, at 2-5, and in the hearing on the motion for reconsideration, see Berger ER, Tab 4203, at 43. The district court seems to have been mistaken in believing that the daily time logs offered with the motion for reconsideration were just the handwritten versions of the printed summaries provided with the original petition. It seems to us that the handwritten time logs provide more detailed information than the printed summaries.15
Because we believe the district court abused its discretion in failing to reconsider its initial fee decision in light of the new handwritten time logs provided by the Berger firm, we vacate the firm’s fee award and remand for reconsideration of its fee request. Our holding that the district court abused its discretion is based only on the narrow ground that it erroneously believed that the new handwritten records provided no new information. We take no position on whether, or in what amount, the district court should alter its fee award to the Berger firm after reviewing the handwritten records on remand.
X
Conclusion
In sum, we vacate the fee award and remand for reconsideration of three elements in Class Counsel’s fee request: (1) Class Counsel’s request for a risk multiplier to enhance the lodestar; (2) the Gold firm’s request for compensation for delay in payment of fees for attorneys who left the firm before the fee request was filed; and (3) the Berger firm’s request that its fee award be reconsidered in light of the handwritten time logs it supplied with its motion for reconsideration.
. The history of this litigation is recounted in In re Washington Pub. Power Supply Sys. Sec. Litig., 720 F.Supp. 1379 (D.Ariz.1989) [hereinafter WPPSS 7] and In re Washington Pub. Power Supply Sys. Sec.Litig., 779 F.Supp. 1063 (D.Ariz.1990).
. Under the lodestar/multiplier method, the district court first calculates the "lodestar" by multiplying the reasonable hours expended by a reasonable hourly rate. Pennsylvania v. Delaware Valley Citizens' Council for Clean Air, 478 U.S. 546, 565, 106 S.Ct. 3088, 3098, 92 L.Ed.2d 439 (1986). The court may then enhance the lodestar with a "multiplier,” if necessary, to arrive at a reasonable fee. Blum v. Stenson, 465 U.S. 886, 888, 104 S.Ct. 1541, 1543, 79 L.Ed.2d 891 (1984). Under the percentage method, the court simply awards the attorneys a percentage of the fund sufficient to provide plaintiffs’ attorneys with a reasonable fee. Paul, Johnson, Alston & Hunt v. Graulty, 886 F.2d 268, 272 (9th Cir.1989). Whether a court applies the lodestar or the percentage method, "we require only that fee awards in common fund cases be reasonable under the circumstances.” Florida v. Dunne, 915 F.2d 542, 545 (9th Cir.1990) (emphasis added). Because a reasonable fee award is the hallmark of common fund cases, and because arbitrary, *1295and thus unreasonable, fee awards are to be avoided, neither method should be applied in a formulaic or mechanical fashion.
. The D.C. Circuit also requires the use of the percentage method in common fund cases. See Swedish Hosp. Corp. v. Shalala, 1 F.3d 1261, 1271 (D.C.Cir.1993). However, other circuits leave it to the discretion of the district court to choose between the lodestar and percentage methods in common fund cases. See, e.g., Rawlings v. Prudential-Bache Properties, Inc., 9 F.3d 513, 516 (6th Cir.1993); Longden v. Sunderman, 979 F.2d 1095, 1099 (5th Cir.1992); Harman v. Lyphomed, Inc., 945 F.2d 969, 975 (7th Cir.1991); Brown v. Phillips Petroleum Co., 838 F.2d 451, 454 (10th Cir.1988); In re "Agent Orange" Prod.Liab.Litig., 818 F.2d 226, 232 (2d Cir.1987).
. The attorneys in Graulty had a contingent fee arrangement with some of the plaintiffs (the estates receiving 30 percent of the common fund), but had no fee arrangement with the plaintiffs receiving the remaining 70 percent of the fund. 886 F.2d at 269-70. Because the attorneys’ efforts benefitted both groups of plaintiffs simultaneously, it would have been nearly impossible to allot the hours spent working on behalf of each group. Id. at 272.
. While the task of calculating the lodestar in this case was immense, it plainly did not tax the district court beyond its limits. In fact, the district court provided an extraordinarily comprehensive and detailed explanation of its lodestar award in its exhaustive 167-page order. See WPPSS II, 779 F.Supp. at 1063-1230.
. On reconsideration, Class Counsel stated that their fee calculation under the "alternative” percentage method was intended to "test the Kerr [lodestar/multiplier] analysis and to confirm the reasonableness of the proposed multipliers.” Reply Memorandum of Class Counsel in Support of Reconsideration on Attorneys’ Fee Order at 7.
. Because Dague was decided in 1992, well after the district court's two 1990 attorneys' fee rulings, the district court had no opportunity to consider the effect of Dague on the use of risk multipliers in common fund cases.
. As yet, there is no consensus whether Dague applies to common fund cases, although most courts have refused to extend Dague's rationale to common fund cases. Compare Edelman v. PSI Assocs. II, Inc., 147 F.R.D. 217, 219 (C.D.Cal.1993) {“Dague and the line of decisions preceding it, by their terms, apply only to 'fee shifting’ cases”) (emphasis added); Rawlings, 9 F.3d at 517 (post-Dague case stating that, in the Sixth Circuit, contingency risk is one factor that is "the focus of a district court’s analysis when determining whether to utilize a multiplier"); Longden v. Sunderman, 979 F.2d 1095, 1099 (5th Cir.1992) (post-Dague case noting that Fifth Circuit still uses multipliers, including for risk, to adjust the lodestar amount); Gottlieb v. Wiles, 150 F.R.D. 174, 185 & n. 6 (D.Colo.1993) (rejecting argument that Dague precludes use of risk multiplier in common fund cases); In re Avon Prods., Inc. Sec. Litig. [1992 Transfer Binder] Fed.Sec.L.Rep. (CCH) ¶ 97,061, at 94, 701, 1992 WL 349768 (S.D.N.Y. Nov. 5, 1992) (post-Dague decision using percentage method, but stating that risk multiplier of 2 to 3 would be appropriate under lodestar method), with Nensel v. Peoples Heritage Fin. Group, Inc., 815 F.Supp. 26, 30 (D.Me.1993) (holding that Dague's proscription of the use of risk multipliers applies to common fund cases); In re Bolar Pharmaceutical Co. Sec-Litig., 800 F.Supp. 1091, 1095 (E.D.N.Y.1992) (same); In re Nineteen Appeals Arising Out of San Juan Dupont Plaza Hotel Fire Litig., 982 F.2d 603, 619 (1st Cir.1992) (Lay, J., concurring) (same).
. See, e.g., Harman, 945 F.2d at 975-76; Brown, 838 F.2d at 454-56; In re “Agent Orange", 818 F.2d at 234 n. 2, 236 (characterizing "the risk-of-success factor as 'perhaps the foremost' factor to be considered” when calculating a reasonable fee in common fund cases).
. In Skelton, the Seventh Circuit also dispelled a related concern, not mentioned explicitly in Dague, about awarding risk multipliers in fee-shifting cases: that "risk multipliers tend to penalize the parties with the strongest defenses. The stronger the defense, the higher the risk involved in bringing the suit and the greater the multiplier necessary to compensate plaintiff's attorney for bringing the action. Thus, defendants with better cases pay higher plaintiffs attorney fees." 860 F.2d at 253. Again, this is not a concern in common fund cases since the burden of compensating class counsel falls on the plaintiff class, not the defendants. Id.
. The abuse of discretion standard "applies not only to the basic fee but also to multipliers.” Chalmers v. City of Los Angeles, 796 F.2d 1205, 1210 (9th Cir.1986) (internal quotation marks omitted).
. The district court itself acknowledged in its initial fee decision that the lodestar rates were "the customary fees charged to fee paying clients.” WPPSS II, 779 F.Supp. at 1099.
. These enhancements increased the overall lodestar from $27 million to $32.46 million, yielding a "blended” multiplier of 1.2.
. This is significant given that over 36 percent of the hours expended by the Gold firm are attributable to attorneys who had left-prior to the filing of the fee petition. Gold Firm’s Opening Br. at 6 n. 5.
. The Berger firm also offered to provide the court with a typed copy of the handwritten records if the court found them illegible. See Berger ER, Tab 4155, at 7.