OPINION
PAEZ, Circuit Judge.
Plaintiffs, who are independent insurance agents for Equitable Life Assurance Society of the United States (“Equitable”), initially brought this class action in California Superior Court challenging Equitable’s modification to the agents’ commission payment system and to certain agents’ health benefits plan. Equitable removed the case to federal court on the grounds that the health benefits claim was completely preempted by the Employee Retirement Income Security Act (“ERISA”), 29 U.S.C. §§ 1001-1461, and that there was diversity jurisdiction. The district court denied Plaintiffs’ motion to remand the action to state court, concluding that there was federal question jurisdiction under ERISA. Soon thereafter, Equitable agreed to settle Plaintiffs’ claim regarding the commission payment system.
As part of the settlement, Equitable agreed to pay Plaintiffs approximately $20 million in commission payments. Having secured a settlement fund, or “common fund,” Plaintiffs’ counsel sought an award of attorney’s fees and costs. Although Plaintiffs’ counsel initially sought 25 percent of the fund and later 10 percent, they were awarded approximately 3 percent of the fund. In arriving at its award, the district court declined to calculate its award on a percentage-of-the-fund approach and instead utilized the lodestar method. Dissatisfied with the result, Plaintiffs and their counsel appealed the fee award.
Before reaching the merits of the appeal, we must first address whether Plaintiffs may attack subject matter jurisdiction in this collateral proceeding. We conclude that they cannot because they litigated the matter before the district court and then relinquished their right to appeal by settling. That decision is now final, and we cannot revisit it here. Accordingly, we do not decide whether the district court erred by denying Plaintiffs’ remand motion on the basis of its determination that it had jurisdiction under ERISA.1
Turning to the attorney’s fee award, we find that the district court did not abuse its discretion by calculating the award according to the lodestar method rather than the percentage-of-the-fund method. The district court erred, however, in its analysis of whether Plaintiffs’ counsel was entitled to a risk multiplier. The district court [1002]*1002also did not adequately explain whether it compensated Plaintiffs’ counsel for the delay in payment. Accordingly, we affirm in part, reverse in part, and remand.
I. Factual and Procedural History
A. Prelitigation Events
In 1992, Equitable announced that it would begin calculating insurance agents’ future commissions on medical policies based on the 1992 policy premium base rather than on future premium increases. Perceiving such action to be a breach of Equitable’s contract with its insurance agents,'agent Michael Siefe complained to Equitable in October 1992. His complaints were to no avail.
Siefe contacted two attorneys, neither of whom was interested in representing him on a eontingeiicy fee basis because (1) his claim was not “clear cut,” (2) his claim only posed “a modest monetary threat (by Equitable’s standards),” (3) Equitable was a large insurer with significant resources to retain highly qualified attorneys, (4) the attorneys were busy with other cases with a better risk/rate-of-return ratio, and (5) the attorneys “were not willing to invest the necessary time” to learn the issues.
Sometime later, in March 1996, Siefe contacted Herbert Adelman, a lawyer; in Washington, D.C. Adelman previously had filed a class action against Equitable on behalf of medical policyholders. After Adelman “determined that there was sufficient merit in the claims to warrant a meeting,” they met for three days in May 1996 to discuss the case. Adelman ultimately agreed to represent Siefe and other similarly situated agents.
Adelman had difficulty securing local counsel in California. He contacted approximately six law firms in California, none of which was interested in the case because (1) the litigation was anticipated to be protracted and expensive, and involved a defendant with great resources, (2) the risk of loss was significant, and (3) there was a possibility of a statute of limitations defense. Adelman also spoke to lawyers in Florida, Texas, and New York firms but could not find anyone interested in filing a case in one of those states. Eventually Adelman found local counsel in California.
On September 19, 1996, Adelman delivered a letter to Equitable’s counsel stating that an association composed of current and former Equitable insurance agents was planning to circulate a letter to all of Equitable’s agents setting forth claims against Equitable unless Equitable was willing to enter into discussions regarding the agents’ claims. Equitable requested additional time as well as documentation to review the claims, but made no commitment regarding how it would respond. Equitable also requested that Adelman not circulate the letter pending its review of the claims.
B. The Lawsuit and the Subsequent Settlement
On October 21, 1996, while Equitable was reviewing the agents’ claims, Plaintiffs filed this class action in California Superior Court alleging two claims for breach of contract.2 The first claim (“Count One”) related to Equitable’s freeze of the premium base for calculating insurance agents’ commissions on medical policies. The second claim (“Count Two”) involved Equitable’s alleged breach of its promise to provide health benefits to certain agents.
Equitable removed the action to federal court on the grounds that Count Two was completely preempted by ERISA and that there was diversity jurisdiction. Plaintiffs responded with a motion to remand the case to state court. Concluding that it had [1003]*1003federal question jurisdiction, the district court denied the motion to remand.
On November 6, 1996, Equitable’s counsel informed Plaintiffs that Equitable was “concerned” about Count One and might agree to settle it. It was not until February 5, 1997, however, approximately three and one-half months after the lawsuit was filed and four and one-half months after negotiations began, that Equitable announced that it would settle Count One by ending the freeze.
C. The June 10, 1997 Order
In response to Equitable’s announcement, Plaintiffs filed a motion for a preliminary injunction to compel Equitable to withhold 25 percent of the settlement fund for attorney’s fees. Concluding that a 10 percent withholding would be sufficient, the district court granted the preliminary injunction in part on June 10, 1997. In making this preliminary determination, the district court explained that it was exercising its discretion to apply the lodestar method, rather than the percentage-of-the-fund method, to calculate attorney’s fees. The court emphasized that the “early settlement ... renders a twenty-five percent recovery for the attorneys unreasonable in light of the circumstances” and that the lodestar approach would avoid a “windfall” to the attorneys at the expense of their clients. The court further explained that “[tjhere has been no discovery, no lengthy settlement negotiations, no protracted litigation of any kind.” The district court then determined that when it ultimately ruled on counsel’s fee application, it would use a “generous” combined hourly rate for partners, associates, and paralegals of $300 per hour.
In discussing whether it would enhance the lodestar amount with a multiplier, the district court stated that it would consider a multiplier of up to 1.5 if Plaintiffs’ counsel secured a 100 percent recovery for the class. The court also explained that it had discretion to apply a risk multiplier, because the attorneys took the case on a contingent basis, and would do so if Plaintiffs’ counsel were to show that (1) without an adjustment for risk they would have had “substantial difficulties” finding counsel, and (2) the difficulty of finding counsel exists for the entire class of contingency fee cases and not just for this particular case. See Fadhl v. City & County of San Francisco, 859 F.2d 649, 650 (9th Cir.1988) (per curiam) (setting forth the two-pronged standard for enhancing a fee award to account for risk). The district court then tentatively ruled that Plaintiffs’ counsel had failed to make the requisite showing to support application of a risk enhancement. Citing the briefs filed in support of the motion for a preliminary injunction, the district court noted that Plaintiffs’ counsel “concede[d] that success on count one of the complaint was likely and that the most substantial risk they faced was that they would only obtain a partial recovery.”3 Nonetheless, the court did not rule out the possibility that Plaintiffs’ counsel could demonstrate that they were entitled to a fee enhancement on the basis of risk or “undesirability.” The court stated, however, that the multiplier would be no greater than two.
Turning to Plaintiffs’ counsel’s request for an enhancement due to the potential delay in payment of their fee award, the district court concluded that no increase would be necessary. First, the court reasoned that there would be no delay in payment because fees derived from retroactive commission payments were [1004]*1004“immediately deductible” once the court determined the amount of the fee award. Second, Equitable agents would be required to pay a “prime rate enhancement” to compensate counsel for the delay in payment for attorney’s fees derived from future payments to Equitable agents.
Plaintiffs filed an interlocutory appeal of the district court ruling on the preliminary injunction motion, which we affirmed. Fischel v. Equitable Life Assurance Soc’y, 127 F.3d 1104, 1997 WL 664934 (9th Cir.1997) (table). We held that the district court did not abuse its discretion by denying the preliminary injunction in part.
D.Further Negotiations and Settlement
After Equitable announced its decision in February 1997 to settle Count One, the parties engaged in extensive negotiations over issues such as interest, benefits, and computations. The parties reached a full settlement approximately one year later.
The district court preliminarily approved the settlement (including certification of a class for purposes of the settlement) and, after class notice and a fairness hearing, entered a stipulated judgment. The amount of the settlement was roughly $18,475,500, which, with accumulated interest, increased to approximately $19,731,690 by September 2000.
The class notice stated that Plaintiffs’ counsel would seek 10 percent of the funds recovered as attorney’s fees. No class member filed an objection to that amount. Plaintiffs’ counsel mailed a survey to each class member (approximately 6,500 agents) seeking feedback on the requested fee. Roughly 1,179 class members responded, supporting an average fee of 8.6 percent and a median fee of 10 percent.
E. The September 3,1999 Order
After entry of the judgment, Plaintiffs’ counsel applied for an award of attorney’s fees and costs, requesting 10 percent, or $1,847,500, of the settlement fund. On September 3, 1999, the district court again denied counsel’s request to calculate fees by the percentage-of-the-fund method, and reaffirmed its earlier decision to calculate fees under the lodestar method.
The court then determined the hours for which Plaintiffs’ counsel were to be compensated. These findings are undisputed. With respect to the hourly fee, the court reaffirmed its determination of a “generous” combined hourly rate of $300.
After noting that the use of a multiplier is the exception rather than the rule, the district court awarded a 1.5 multiplier for counsel’s 100 percent success rate. The court declined, however, to apply a multiplier for the contingent nature of the case and the risk of nonpayment. Although Plaintiffs’ counsel argued that it took years for Plaintiffs to obtain counsel and that Equitable was a “tough adversary,” both of which increased the “undesirability” of the case, the court restated its reasoning in the June 1997 order and refused to apply a risk multiplier. The district court also declined to adjust the attorney’s fee award for any delay in payment.
After applying the 1.5 multiplier, the court awarded $572,413.50 in attorney’s fees, plus expenses, for a total award of $686,226.25, or approximately 3 percent of the settlement fund.
F. The May 5, 2000 Order
Plaintiffs’ counsel moved to amend the September 1999 order, requesting, among other things, interest on the amount awarded to compensate for the delay in payment. Although the district court declined to award interest, it amended the [1005]*1005payment plan that it had earlier established. The court determined that the entire attorney’s fee award should be deducted from the initial settlement fund rather than from both the initial payments to agents as well as future payments. This change in payment method, the court concluded, “effectively moot[ed] class' counsel’s motion for interest on the portion of its fees that under the court’s prior order were to be paid in the future.” The court explained that there was no need to compensate counsel for the delay in payment because the “generous hourly rate applied in calculating the fee award sufficiently compensates counsel for delay between the date of settlement and the award of fees and represents counsel[’]s current rates.” Plaintiffs and their counsel timely appealed from this final order regarding attorney’s fees and costs.4
II. Standard of Review
We review an attorney’s fee award for an abuse of discretion. Fischer v. SJB-P.D., Inc., 214 F.3d 1115, 1118 (9th Cir.2000). “A district court abuses its discretion if its decision is based on an erroneous conclusion of law or if the record contains no evidence on which it rationally could have based its decision.” Paul, Johnson, Alston & Hunt v. Graulty, 886 F.2d 268, 270 (9th Cir.1989). We review the underlying factual determinations for clear error and review de novo any legal analysis relevant to the fee determination. Fischer, 214 F.3d at 1118.
III. Discussion
A. Relitigation of Subject Matter Jurisdiction
Plaintiffs contend that we lack jurisdiction over this action because the district court erroneously concluded that Count Two is completely preempted by ERISA. Equitable argues that Plaintiffs never appealed the judgment, and thus res judicata precludes them from challenging subject matter jurisdiction in this appeal of the attorney’s fee award, a collateral post-judgment proceeding. Plaintiffs counter that they could not have appealed the judgment because they received all the relief that they sought. Because they could not appeal, Plaintiffs contend that they should not be collaterally estopped— the applicable res judicata doctrine — from raising the jurisdictional issue here.5
We recently addressed whether a party could relitigate in an attorney’s fee proceeding the determination of subject matter jurisdiction in the underlying action. See Zambrano v. INS, 282 F.3d 1145, 1151 (9th Cir.2002), as amended, 00-16191, 302 F.3d 909, 2002 WL 2012583 (Sept. 4, 2002). We explained that the district court had “fully decided the issue of jurisdiction, and that judgment has now become final.” Id. Accordingly, we held, consistent with “tra[1006]*1006ditional notions of issue and claim preclusion,” that the plaintiffs could not relitigate jurisdiction in the fee proceedings. Id; see also Carpenters S. Cal. Admin. Corp. v. Knight (In re Knight), 207 F.3d 1115, 1116, 1117 (9th Cir.2000) (holding that, in the attorney’s fee proceeding, we must accept as true the district court’s determination that it lacked subject matter jurisdiction over the case because the plaintiff did not appeal the dismissal).
Similarly here, Plaintiffs cannot reliti-gate subject matter jurisdiction in the fee proceeding. Plaintiffs had a full and fair opportunity to litigate jurisdiction before the district court. That they could not appeal does not change our conclusion. They agreed to settle Count One and consented to entry of a stipulated judgment on that claim, thus “voluntarily surrendering] their right to further review[.]” Greenleaf v. Garlock, Inc., 174 F.3d 352, 359 (3rd Cir.1999). Indeed, the terms of the settlement were consistent with the district court’s jurisdictional ruling. By their voluntary settlement, Plaintiffs rendered the district court’s jurisdictional ruling unreviewable. Because that ruling is final, Plaintiffs cannot now challenge subject matter jurisdiction. We therefore turn to the dispute over the attorney’s fee award.
B. Attorney’s Fee Award
Plaintiffs’ counsel dispute the district court’s fee award, arguing that (1) the fee award is unreasonably low, (2) the court abused its discretion by declining to apply a risk multiplier, and (3) the court erred by failing to award interest on the fee award to account for delay in- payment.6 Although we find that the fee award was not unreasonably low, we conclude that the district court erred in its analysis of the risk and delay factors.
1. Unreasonably Low Fee Award
When counsel recover a common fund which confers a “substantial benefit” upon a class of beneficiaries, they are entitled to recover their attorney’s fees from the fund. Lewis v. Anderson, 692 F.2d 1267, 1270 (9th Cir.1982). This is a “well-recognized exception” to the general rule that parties are responsible for their own attorney’s fees. Boeing Co. v. Van Gemert, 444 U.S. 472, 478, 100 S.Ct. 745, 62 L.Ed.2d 676 (1980). Because the attorneys seek fees on their own behalf, which may be contrary to their clients’ interests, the district court must “look out for the interests of the beneficiaries, to make sure that they obtain sufficient financial benefit after the lawyers are paid.” Florida ex rel. Butterworth v. Exxon Corp. (In re Coordinated Pre-trial Proceedings in Petroleum Prods. Antitrust Litig.), 109 F.3d 602, 608 (9th Cir.1997) (“Coordinated Pretrial ”).
In a common fund case, the district court has discretion to apply either the lodestar method or the percentage-of-the-fund method in calculating a fee award. Vizcaino v. Microsoft Corp., 290 F.3d 1043, 1047 (9th Cir.2002), petition for cert. filed 71 U.S.L.W 3154 (U.S. Aug. 13, 2002) (No. 02-252); Coordinated Pretrial, 109 F.3d at 607. We have established a 25 percent “benchmark” in percentage-of-the-fund cases that can be “adjusted upward or downward to account for any unusual circumstances involved in [the] case.” Paul, Johnson, Alston & Hunt, 886 F.2d at 272. Under the lodestar method, the court multiplies a reasonable number of hours by a reasonable hourly rate. Class [1007]*1007Plaintiffs v. City of Seattle (In re Wash. Pub. Power Supply Sys. Sec. Litig.), 19 F.3d 1291, 1294 n. 2 (9th Cir.1994) (“Washington Public ”).7 There is a “strong presumption” that the lodestar figure represents a reasonable fee. D’Emanuele v. Montgomery Ward & Co., 904 F.2d 1379, 1384 (9th Cir.1990), overruled on other grounds by Burlington v. Dague, 505 U.S. 557, 112 S.Ct. 2638, 120 L.Ed.2d 449 (1992). Thus, although a court can adjust the lodestar upward or downward based on certain factors, adjustments are “the exception rather than the rule.” Id. at 1383-84.
“Reasonableness is the goal, and mechanical or formulaic application of either method, where it yields an unreasonable result, can be an abuse of discretion.” Coordinated Pretrial, 109 F.3d at 607; see also Paul, Johnson, Alston & Hunt, 886 F.2d at 271 (holding that attorney’s fee award in a common fund case must be “ ‘reasonable’ under the circumstances”). When using the percentage-of-the-fund approach, although there is a 25 percent benchmark, that rate may be unreasonable in some cases. Vizcaino, 290 F.3d at 1047-48; see also Washington Public, 19 F.3d at 1297-98. Courts may compare the two methods of calculating attorney’s fees in determining whether fees are reasonable. See Coordinated Pretrial, 109 F.3d at 607.
Because district courts have the discretion to calculate attorney’s fees by either the lodestar or percentage-of-the-fund approach, the district court here did not abuse its discretion by using the lodestar method to calculate fees. The fact that the case was settled early in the litigation supports the district court’s ruling; the 25 percent benchmark of the percentage-of-the-fund approach might very well have been a “windfall.” Nor did the district court err by failing to compare the lodestar result to the 25 percent benchmark. Additionally, despite Plaintiffs’ counsel’s assertion that they should have been compensated for the size of the fund they obtained, the district court compensated counsel for this achievement when the district court applied a 1.5 multiplier for their 100 percent success rate.
Further, we are not persuaded that the district court abused its discretion by failing to award 10 percent of the fund when no class member objected to that percentage. Plaintiffs’ counsel cite Gunter v. Ridgewood Energy Corp., 223 F.3d 190 (3rd Cir.2000), for the proposition that the district court should have considered this fact. See id. at 195 n. 1; id. at 199 (“[N]o one in the class objected to Counsel’s request for fees. Yet, a client’s views regarding her attorneys’ performance and their request for fees should be considered when determining a fee award.”). A district court should consider “the presence or absence of substantial objections by members of the class to the ... fees requested by counsel,” id. at 195 n. 1, but [1008]*1008this factor is not outcome determinative; it must be considered in light of all of the other factors. Here, the district court did not abuse its discretion by failing to increase the attorney’s fee award to account for the class members’ view of the requested fee award because (1) there was an early settlement, (2) the district court used the lodestar rather than the percentage-of-the-fund approach to calculate fees, and (3) the district court applied a 1.5 multiplier for counsel’s 100 percent success rate.
2. Risk of Nonpayment
“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.” Washington Public, 19 F.3d at 1299; see also Vizcaino, 290 F.3d at 1051. This provides the “necessary incentive” for attorneys to bring actions to protect individual rights and to enforce public policies. In re “Agent Orange” Prod. Liab. Litig., 818 F.2d 226, 236 (2d Cir.1987). In common fund cases, there is no concern about financially burdening a defendant to compensate for the risk of nonpayment, because the attorney’s fee award is deducted from the plaintiffs’ fund. Washington Public, 19 F.3d at 1300. In such cases, the plaintiffs “should share the wealth with the lawyers whose skill and effort helped create it.” Id.
A district court generally has discretion to apply a multiplier to the attorney’s fees calculation to compensate for the risk of nonpayment. Coordinated Pretrial, 109 F.3d at 609. It is an abuse of discretion to fail to apply a risk multiplier, however, when (1) attorneys take a case with the expectation that they will receive a risk enhancement if they prevail, (2) their hourly rate does not reflect that risk, and (3) there is evidence that the case was risky. See Washington Public, 19 F.3d at 1301-02; see also Coordinated Pretrial, 109 F.3d at 609.8
With regard to the first factor, Plaintiffs’ counsel presented evidence that they took this case with the expectation that they would receive a risk enhancement if they prevailed. Adelman stated:
In view of the anticipated vigorous and able defense by Equitable, the absence of a “mega” claim, and Equitable’s refusal over the years to reverse its position, I and my colleagues would not have undertaken this litigation on a contingent fee basis had it been thought that there was any likelihood of a fee being restricted to a small percentage of the amounts recovered. Although I knew that the Court ultimately sets the fee, and that no amount was guaranteed, I was specifically aware in taking the matter on that the usual range of fee awards in common fund cases was 20-30 percent.
[1009]*1009Additionally, Michael Malakoff, another one of Plaintiffs’ attorneys, explained that he
d[id] not believe [that his] law firm would have agreed in advance to undertake a complex class action for an aggregate award for all counsel of less then 20% of a common fund where the fund, as here, was not anticipated to be over $20 million.
In light of the Federal Judicial Center’s report of a mean rate of 27 percent and a median rate of 29 percent for class settlements approved in district courts in Northern California from 1992 to 1994, see Thomas E. Willging et al., Empirical Study of Class Actions in Four Federal District Courts: Final Report to the Advisory Committee on Civil Rules 147 (1996), counsel’s expectation that they would receive a higher percentage of the fund was reasonable. See Washington Public, 19 F.3d at 1302 (holding that counsel’s expectations that they would receive a fee enhancement for risk “appear[ed] to have been reasonable, especially given the common practice in the 1980s for courts to award multipliers in common fund cases”). Plaintiffs’ counsel’s expectation was reasonable because other class action attorneys refused to represent Plaintiffs and Adelman had difficulty finding co-counsel. The district court erroneously ignored these facts in concluding that it would not apply a risk multiplier.
With respect to Plaintiffs’ counsel’s hourly rate, that rate did not reflect any risk of nonpayment. They explained that their “reasonable hourly rate” of $350 for an attorney did “not take into account any contingency enhancement.” The district court acknowledged this fact when it stated, “[i]n the present case, the court adopted hourly rates consistent with the prevailing market in calculating a reasonable fee.”
The final consideration is the risk involved in the case. This includes the strength, both legal and factual, of the case. Although there is no dispute that a court should consider risk at the “outset” of litigation, see, e.g., Florin v. Nationsbank of Georgia, 34 F.3d 560, 565 (7th Cir.1994), courts disagree about exactly when this should be. Compare Goldberger v. Integrated Res., Inc., 209 F.3d 43, 55 (2d Cir.2000) (holding that risk must be measured at the time the lawsuit is filed), with In re Fine Paper Antitrust Litig., 751 F.2d 562, 583 (3rd Cir.1984) (holding that risk is “measured at the point when the attorney’s time was committed to the case”). We hold that risk should be assessed when an attorney determines that there is merit to the client’s claim and elects to pursue the claim on the client’s behalf. This will likely occur before a lawsuit is filed.
In determining risk here, the district court focused on Plaintiffs’ counsel’s statement that they believed that Count One was “meritorious” at the “outset” of the litigation. From this statement, the court inferred that counsel did not assume any risk in pursuing litigation on Plaintiffs’ behalf. It is unclear, however, to what point in time “outset” referred. It could refer to when (1) Adelman began to investigate the ease in early 1996, (2) he determined to pursue the case, (3) Equitable stated that it would review the agents’ claims in September 1996, (4) Plaintiffs’ counsel filed the lawsuit in October 1996, or (5) Equitable announced that it would settle the case in February 1997. Rather than just focusing on counsel’s statement, the court should have considered the entire record to determine when Plaintiffs’ counsel decided to pursue the agents’ claims and should have assessed risk at that time.
[1010]*1010Although the evidence does not demonstrate when Plaintiffs’ counsel decided to represent Plaintiffs, at the very latest counsel had determined to pursue Plaintiffs’ claims when they filed the lawsuit in October 1996. By that date, Equitable had stated that it wanted to review the agents claims, but it had not yet stated that it would consider settling the case. As Defendants had not yet agreed to settle and Plaintiffs’ counsel articulated several concerns that they had about legal issues in the case, such as possible defenses, there appears to have been some risk when the lawsuit was filed.
Accordingly, the evidence demonstrates that the district court failed fully to analyze the risk factor and thus abused its discretion in determining whether to award a risk multiplier. We remand to the district court to review the entire record in order to determine whether to apply a risk multiplier in light of (1) Plaintiffs’ counsel’s expectation that they would receive a risk enhancement, (2) the fact that their hourly rate did not reflect any risk, and (3) the risk of litigation at the time Plaintiffs’ counsel decided to pursue the agents’ claims.
3. Delay in Payment of Attorney’s Fees
Plaintiffs’ counsel contend that, in its May 2000 order determining the final award of attorney’s fees, the district court should have awarded them the 8 percent interest that had accrued on their fees since the court’s June 1997 order directing that 10 percent of the fund be withheld for attorney’s fees.
Attorneys in common fund cases must be compensated for any delay in payment, Coordinated Pretrial, 109 F.3d at 609, and thus Plaintiffs’ counsel are entitled to such compensation. Contrary to counsel’s contention, however, they are not entitled to the 8 percent interest. Instead, the district court had discretion to compensate them either “ ‘(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.’” Id. (quoting Washington Public, 19 F.3d at 1305).
The district court, however, chose neither method to compensate Plaintiffs’ counsel for the delay in payment. The court explained that, after it had determined that all of the attorney’s fees should be deducted from the initial settlement fund, it did not need to compensate counsel for the delay in payment because of the “generous” hourly rate they had received. The district court chose the “generous” rate in June 1997, however, more than two years before the court’s September 1999 order determining the fee award. The “generous” rate is sufficient only if it reflects the September 1999 average rate for partners, associates, and paralegals at Plaintiffs’ counsel’s firm,9 or reflects a prime enhancement.
Thus, we remand for the district court to determine whether the $300 hourly rate represents Plaintiffs’ counsel’s firm’s September 1999 rates or includes a prime enhancement to account for any delay. If the rate does not reflect either of these requirements, then the court must adjust the fee award accordingly.
[1011]*1011With respect to the eight-month delay from September 1999 until May 2000, while Plaintiffs’ motion to amend the September 1999 order was pending, the district court may, in its discretion, compensate for this delay. In making its determination, the district court should consider the length of the delay as well as the amount of the fee award involved. See Barjon v. Dalton, 132 F.3d 496, 502-03 (9th Cir.1997) (holding that the plaintiffs’ counsel were not entitled to compensation for a seventeen-month delay in the payment of approximately $11,500 in attorney’s fees in one matter and a fourteen-month delay in the payment of approximately $8,000 in fees in another matter, because “the delay itself was not very long and the amount of fees not very high”).
IV. Conclusion
For the foregoing reasons, we remand this case to the district court to determine whether to apply a risk multiplier and to determine whether it provided adequate compensation for the delay in payment of attorney’s fees.
AFFIRMED in part, REVERSED in part, and REMANDED for further proceedings consistent with this opinion. The parties shall bear their own costs on appeal.