STEPHEN H. ANDERSON, Circuit Judge.
Thomas H. Connolly, Chapter 7 trustee for the debtor, MiniScribe Corporation, appeals from orders of the district court (1) reversing an order of the bankruptcy court awarding him a trustee’s fee of $3,044,953.69, and (2) ordering that he be allowed a fee of $1,828,812. We affirm.
FACTS
On January 1, 1990, the debtor filed a voluntary petition in bankruptcy under Chapter 11. It converted the case to a Chapter 7 liquidation on April 16, 1991. Ten days later, Connolly was appointed Chapter 7 trustee.
At the time Connolly assumed his duties as trustee, the MiniScribe estate was insolvent. The estate had only $150,000 cash on hand, and faced Chapter 11 administrative expenses exceeding $3.0 million, a su-perpriority claim against it by Standard Chartered Bank (SCB) in the amount of $17 million, and total claims of approximately $900 million. The only opportunities to develop a bankruptcy estate lay in recoveries on avoidance claims and in a fraud action against former MiniScribe officers and outside accountants and consultants.
The bankruptcy court found that Connolly performed admirably in obtaining funds for the estate. He not only convinced SCB to reduce its claim from $17 million to $1 million, but also further persuaded it to loan the estate $1 million to defray the cost of prosecuting its fraud action. This action was eventually settled on terms highly favorable to the bankruptcy estate and its creditors, yielding over $80 million for the payment of claims. Connolly also conducted 45 adversary proceedings that realized over $17 million for the estate and that resulted in the reduction of claims against the estate from approximately $900 million to approximately $168 million. Finally, Connolly settled a recovery action against one of MiniScribe’s creditors by taking an equity interest in the creditor which Connolly ultimately sold for over five times the value of the settlement.
The bankruptcy court made a number of interim fee payments to Connolly as trustee. In his final fee application, he sought additional compensation that would have resulted in a total fee of $3,044,953. This was the maximum fee permitted at that [1238]*1238time under the percentage fee scheme (the “cap”) outlined in 11 U.S.C. § 326(a) (1986).1 Appellee Harris Trust Company, representing MiniScribe’s debenture holders, opposed the request, as did the United States trustee.
1. First bankruptcy court decision
The bankruptcy court carefully analyzed Connolly’s fee request and issued an extensive order and opinion approving the statutory maximum payment. Connolly v. Harris Trust Co. (In re Miniscribe Corp.), 241 B.R. 729 (Bankr.D.Colo.1999). The court determined that during his administration, Connolly had disbursed $101,492,332 through his accounts. This amount, which included settlements of fraud actions that the trustee and other litigants had brought against third party defendants, would be used in calculating the statutory maximum under the “cap” set out in § 326(a).
The bankruptcy court next applied a lodestar analysis, focusing on a reasonable number of hours spent by the trustee on his duties multiplied by a reasonable hourly rate. It noted that Connolly’s time records reflected 1,779 hours spent on the case, along with an additional 1,347 hours spent by his paralegals. The bankruptcy court concluded that this amount of time, however, was misleadingly low because Connolly had performed more efficiently than expected. It rejected the objectors’ position that the hourly rate should be differentiated, with a rate of $1,000 to $1,500 awarded for time spent on the fraud litigation and a more modest rate of $250 awarded for time spent on other activities. The court determined that a uniform, reasonable hourly rate of $500 should be applied to the 1,779 hours reasonably expended by Connolly, resulting in a lodestar fee of $889,500.
The court next adjusted the lodestar fee by considering the factors outlined in Johnson v. Georgia Highway Express, Inc., 488 F.2d 714 (5th Cir.1974). It determined that these factors (novelty and difficulty of the issues; requisite skill; preclusion of other employment; contingent nature of the fee; time limitations; amount involved and results obtained; and the experience, reputation, and ability of the trustee) justified a lodestar multiplier of 3.5, resulting in a total fee of $3,113,250.
Alternatively, the court applied a “common fund” or “percentage of the fund” analysis to arrive at an appropriate fee. If the trustee were allowed his claimed fee of $3,044,954, the total adjusted costs for raising and administering the “common fund” of the bankruptcy estate would be $10,864,402, approximately 10.7 percent of the total funds ($101,492,456) that had been administered by the trustee. This percentage, the bankruptcy court determined, was well within the range of permissible costs on a percentage basis.
2. First district court decision
Harris Trust appealed from the bankruptcy court’s award. In an unpublished order and judgment, the district court rejected application of the common fund analysis under the circumstances of this case. Under a common fund theory, the court reasoned, the entire cost of recovery of the amount disbursed by Connolly would be borne by the only remaining creditors, the subordinated debenture holders. Moreover, Connolly was not solely responsible for creation of the fund; the plaintiffs whose fraud claims had been con[1239]*1239solidated with his own had also been represented by able counsel and the bankruptcy court had not factored in the fees, costs and expenses paid to them.
The district court also rejected the $500 per hour unified lodestar rate as lacking in evidentiary support. It found the 3.5 multiplier unjustified because it overstated the risk of non-payment to Connolly, ignored the routine nature of much of the work he had performed, and overlooked the contribution of the trustee’s counsel and counsel for the other fraud plaintiffs to the recovery of the fund. The bankruptcy court would have to recalculate the adjusted lodestar amount on remand.
Finally, the district court addressed the statutory cap. It found that MiniScribe’s bankruptcy estate had been unusual because of the amount of money passing through the trustee’s accounts and because of the circumstances surrounding those disbursements. It noted that $70 million of the approximately $101 million that had passed through the trustee’s accounts had been paid to other parties who had consolidated their fraud actions with the trustee’s fraud action for purposes of trial. The correct measure of the results achieved for purposes of computing the statutory cap lay somewhere between the $31.5 million contended for by Harris Trust, and the $101.5 million figure adopted by the bankruptcy court. The precise figure was left to be calculated by the bankruptcy court on remand.
3. Second bankruptcy court decision
The bankruptcy court held further hearings on remand and took up the fee issues in a second published decision, Connolly v. Harris Trust Co. (In re MiniScribe Corp.), 257 B.R. 56 (Bankr.D.Colo.2000). It recalculated the size of the estate, leaving a figure of $67.4 million attributable to Connolly’s efforts.
The bankruptcy court did not read the district court’s order to preclude application of a common fund analysis; it believed the district court had only rejected its application based on the bankruptcy court’s prior findings. Therefore, having adjusted these findings, the bankruptcy court once again applied a common fund percentage analysis to the $67.4 million estate realized by Connolly’s efforts. Assuming Connolly were awarded the $3.04 million he sought, the $10.86 million total administrative burden would still be within a reasonable range under a common fund analysis.
The bankruptcy court next turned to the lodestar analysis. Addressing the district court’s criticism of its previous effort, it conceded that there was no direct support for the $500 per hour figure in the record. Upon consideration of Connolly’s highly responsible role and the fees charged by other professionals, however, the bankruptcy court determined that while a fee of $500 per hour would not be unreasonable, a rate of $400 per hour would also be reasonable. Applied to the 1,779 hours accounted for by Connolly, the base lodestar amount would be $711,600.
The bankruptcy court next turned to the district court’s criticism of its use of a 3.5 lodestar multiplier. After again analyzing the trustee’s contribution to and success achieved in the case, it concluded that a multiplier of 2.57 would be appropriate. Thus, the adjusted lodestar fee would be $1,828,812. Once again, however, the bankruptcy court concluded that the lodestar fee was not the appropriate measure of reasonable compensation; instead, it allowed a larger amount of $3,044,953.69, calculated under a common fund approach.
4. Second district court decision
Harris Trust again appealed from the bankruptcy court’s award. The district [1240]*1240court again concluded that a common fund analysis was inappropriate in this case. It found that there was no way to quantify Connolly’s contribution to the bankruptcy estate that would be comparable to the fund creation theory supporting application of the common fund method. The district court opined that the lodestar method, rather than the common fund method, was the appropriate measure of compensation for this case. It concluded that the adjusted lodestar fee of $1,828,812 arrived at by the bankruptcy court in its second decision constituted a reasonable fee within its sound discretion.
ANALYSIS
1. Standard of review
The district court’s first order was not a final, appealable decision under 28 U.S.C. § 158(d), because it remanded the case to the bankruptcy court for “significant further proceedings” of a discretionary nature. Office of Thrift Supervision v. Overland Park Fin. Corp. (In re Overland Park Fin. Corp.), 236 F.3d 1246, 1251 (10th Cir.2001). Connolly’s appeal from the district court’s second, final order, however, permits us to address issues pertaining to both orders, as he requests. See Masunaga v. Stoltenberg (In re Rex Montis Silver Co.), 87 F.3d 435, 437-38 (10th Cir.1996). Conoco, Inc. v. Styler (In re Peterson Distrib., Inc.), 82 F.3d 956, 959 (10th Cir.1996) (citations omitted).
Our review of the bankruptcy court’s decision is governed by the same standards of review that govern the district court’s review of the bankruptcy court. Accordingly we review the bankruptcy court’s legal determinations de novo and its factual findings under the clearly erroneous standard. A finding of fact is clearly erroneous if it is without factual support in the record or if, after reviewing all of the evidence, we are left with the definite and firm conviction that a mistake has been made.
Seeking to reap the deferential standard of review accorded to the bankruptcy court’s factual findings favorable to his position, Connolly paints all the underlying issues here as factual. He contends that the district court improperly supplanted the bankruptcy court’s factual findings concerning the comparable non-bankruptcy cost of his services, a reasonable hourly rate for his services, and the contingent nature of his fee. Hams Trust, by contrast, argues that this case also presents an important legal issue requiring our de novo review. It asks whether a bankruptcy trustee who does not act as his own litigation counsel may ever be compensated based on a percentage-of-the-fund (common fund) methodology. We will exercise de novo review to conclude that under the applicable statutes common fund methodology is not an appropriate form of compensation for a Chapter 7 trustee.
2. Statutory authority
Authority for payment of a reasonable fee to a Chapter 7 trustee is found in 11 U.S.C. §§ 326 and 330. Section 330(a) provides that a bankruptcy court may award a trustee
(1) reasonable compensation for actual, necessary services rendered by such trustee ... and by any paraprofessional persons employed by such trustee ... based on the nature, the extent, and the value of such services, the time spent on such services, and the cost of comparable services other than in a case under this title; and
(2) reimbursement for actual, necessary expenses. .
Section 326(a) provides a formula that limits the compensation of a Chapter 7 trustee arrived at under § 330:
[1241]*1241In a case under chapter 7 ... the court may allow reasonable compensation under section 330 of this title of the trustee for the trustee’s services, payable after the trustee renders such services, not to exceed fifteen percent on the first $1,000 or less, six percent on any amount in excess of $1,000 but not in excess of $3,000, and three percent on any amount in excess of $3,000, upon all moneys disbursed or turned over in the case by the trustee to parties in interest, excluding the debtor, but including holders of secured claims.
Under this standard, § 326(a) sets the maximum compensation payable to the trustee; it does not establish a presumptive or minimum compensation.
The computation in [§ 326(a)] is a limitation on compensation, not a mandate for minimum commissions. The section states that the court may allow only “reasonable compensation under section 330.” Thus, the cap of section 326(a) is implicated only when the compensation is reasonable, and reasonableness is a determination that must begin with 11 U.S.C. § 330.
In re Butts, 281 B.R. 176, 178 (Bankr.W.D.N.Y.2002).
Accordingly, a court awarding trustee fees must begin by assessing reasonableness under § 330(a) before applying the percentage-based cap under § 326(a).
3. Common fund theory
The common fund doctrine was first recognized by the Supreme Court in Trustees v. Greenough, 105 U.S. 527, 26 L.Ed. 1157 (1881), and most recently discussed in Boeing Co. v. Van Gemert, 444 U.S. 472, 100 S.Ct. 745, 62 L.Ed.2d 676 (1980). Essentially, the doctrine holds that “a litigant or lawyer who recovers a common fund for the benefit of persons other than himself or his client is entitled to a reasonable attorney’s fee from the fund as a whole.” Boeing, 444 U.S. at 478, 100 S.Ct. at 749. It is most appropriate when (1) the class of persons benefitted by the lawsuit is small and easily identifiable, (2) the benefits can be traced with some accuracy, and (3) the costs of the litigation can be shifted accurately to those who profit by it. This is an equitable doctrine, and is usually justified on the ground that it prevents unjust enrichment of the non-litigants, who have taken a free ride on the trailblazer’s efforts. Jurisdiction over the fund enables a court to prevent such inequity by assessing fees out of the entire fund, spreading the burden proportionately among those benefitted.
Matter of Fesco Plastics Corp., 996 F.2d 152, 157 (7th Cir.1993) (further citations omitted).
As Connolly notes, we have approved the use of common fund methodology to award attorneys whose efforts created a fund benefiting persons other than themselves or their clients. See, e.g., Gottlieb v. Barry, 43 F.3d 474, 487-88 (10th Cir.1994). This case, however, presents a different issue: whether a percentage-based approach analogous to the equitable doctrine of common fund has a place in establishing “reasonable” statutory compensation for a Chapter 7 trustee.
The available authority on this question is quite limited. What authority there is divides sharply on the question. Recognizing this fact, a national report on professional compensation in bankruptcy cases takes an agnostic approach to the question:
[I]t appears that a majority [of courts] employ the lodestar approach to trustees’ fee determinations. Some courts, however, reject that approach as inappropriate in light of the significant difference between the functions per[1242]*1242formed by trustees and those performed by the professionals employed by trustees. In these courts especially, the results obtained in the case play a far greater role in determining reasonable compensation for trustees than they do with respect to the compensation of professionals.
Am. Bankr.Inst., American Bankruptcy Institute National Report on Professional Compensation in Bankruptcy Cases (G.R. Warner rep.1991), at 206.
We have found no cases in this circuit that discuss whether a common fund or other percentage-based rationale may be used in setting trustee compensation under § 330. Connolly cites a case specifically endorsing a common fund or percentage-based approach for trustee compensation, In re Guyana Development Corp., 201 B.R. 462 (Bankr.S.D.Tex.1996). Harris Trust responds with a case in which a trustee’s common fund argument was rebuffed, In re Marvel Entertainment Group, 234 B.R. 21 (Bankr.D.Del.1999).
In Guyana Development, the bankruptcy court rejected a lodestar approach and awarded the trustee a percentage based fee analogous to a common fund recovery, measured by the maximum permitted under 11 U.S.C. § 326. The methodology employed in Guyana Development has been seriously criticized, however, because the court applied a “sliding scale” approach that essentially incorporated the maximum permitted compensation (calculated on a percentage basis by statute) into the calculation of a “reasonable See.” See In re Neill, 242 B.R. 685, 689 (Bankr.D.N.D.1999); Marvel Entertainment, 234 B.R. at 38-39. For this reason, we do not find Guyana Development persuasive.
In Marvel Entertainment, by contrast, the district court rejected a trustee’s request for a percentage-based fee. In its well-reasoned decision, the court gave several reasons for rejecting the request, many of which are also applicable here: (1) section 326 serves only as a cap and does not establish the trustee’s entitlement to a commission or percentage of amounts disbursed from the estate; (2) there is no principled relationship between the amounts disbursed by a debtor corporation and reasonable compensation based on the trustee’s actual efforts; (3) there is no support in case law for a common fund approach; (4) setting reasonable compensation based on a percentage'of the debt- or’s disbursements would create a substantial risk of abuse in the selection and appointment of trustees by the United States trustee; and (5) a percentage compensation approach could discourage courts from appointing a trustee when it might otherwise be appropriate. Marvel Entertainment, 234 B.R. at 38-40.
Turning to the statutory language, § 330 requires a court awarding trustee’s fees to consider “the nature, the extent, and the value of such services, the time spent on such services, and the cost of comparable services other than in a case under this title.” 11 U.S.C. § 330(a)(1). The “cost of comparable services” factor plays an “overarching role” in assessing the reasonableness of a trustee’s fee, given the nature and extent of the services rendered. Staiano v. Cain (In re Lan Assoc. XI, L.P.), 192 F.3d 109, 124 n. 9 (3d Cir.1999).
Connolly argues that the “cost of comparable services” factor should take into account the fact that attorneys can and do receive a common fund or percentage-based recovery in non-bankruptcy cases. See, e.g., Brown v. Phillips Petroleum Co., 838 F.2d 451, 454 (10th Cir.1988). This argument, however, begs an essential question: whether the services provided by a trustee are “comparable” to those provided by an attorney in such cases. It is far from clear that they are. Trustees [1243]*1243do not have to be attorneys; indeed, the two serve considerably different functions. See Guyana, 201 B.R. at 479 & n. 11. An attorney who serves as a trustee, in fact, may not be paid attorney’s fees for performing duties that properly belong to his role as trustee. 11 U.S.C. § 328(b).
Where the “common fund” has been created through litigation, the distinction is particularly apt. An attorney litigating a case to judgment supplies a significantly different expertise and' function than that of a trustee. The contingent fee trial lawyer generally advances his own money to finance litigation. He enters his appearance before a court and does every task that the litigation requires, personally or through other attorneys he employs, from drafting the complaint to making the closing argument. Having entered his appearance, he has no right to withdraw and his time commitment is not limited or flexible. The trustee, by contrast, serves as a fiduciary whose non-bankruptcy counterpart is a receiver, an executive or an in-house counsel. He is the litigator’s client and has a right to periodic compensation, such as Connolly received here.
Where, as here, the fund was created through negotiation by the trustee in connection with the litigation, there may be more overlap between the functions of trustee and attorney. Nevertheless, the overriding differences between the functions of an attorney and a trustee lead us to conclude that cases in which an attorney has been paid on a percentage basis for creating a common fund do not represent the “comparable cost” of a trustee’s services, even where those services may have contributed to the creation of the fund.
Section 330 also requires the court to consider “the time spent on such services.” If reasonable compensation for a trustee were predicated solely on a common fund recovery, this factor could improperly be shortchanged.2
While it cannot be denied that results obtained is a factor to be considered in assessing the reasonableness of trustee compensation, see, e.g., In re Rauch, 110 B.R. 467, 473-74 (Bankr.E.D.Cal.1990), we believe this factor is bet ter considered as merely one of the Johnson criteria for determining the multiplier, if any, to be applied to the lodestar amount, rather than the sine qua non of the reasonableness calculation.
For all the foregoing reasons, we reject a “common fund” or percentage-based rationale for calculating reasonable trustee compensation under § 330.
4. Lodestar calculation
Having determined that percentage-based compensation is unavailable under § 330, we turn to the appropriate basis for calculation of the fee. We conclude that the lodestar test, with appropriate enhancements under Johnson, is the appropriate method of calculation. This method encompasses each of the factors identified in § 330(a), as well as other factors that may be relevant in calculating a reasonable fee.
The lodestar test is presently used as the dominant method for assessing fees in fee-shifting disputes in federal court. See generally, Gisbrecht v. Barnhart, 535 U.S. 789, -, 122 S.Ct. 1817, 1825, 152 L.Ed.2d 996 (2002). Strictly speaking, [1244]*1244trustee’s fees in bankruptcy do not involve a fee-shifting rationale. This court has long applied the Johnson lodestar factors to assess “reasonableness” of attorney’s fees in a variety of contexts,'however, and has also specifically determined that the test applies to attorney fee determinations under § 330(a)(1). Rubner & Kutner, P.C. v. United States Trustee (In re Lederman Enters., Inc.), 997 F.2d 1321, 1323 (10th Cir.1993) (attorney’s fees); see also generally First Nat’l Bank of Lea County v. Niccum (In re Permian Anchor Servs., Inc.), 649 F.2d 763, 768 (10th Cir.1981). Other courts have applied an adjusted lodestar test, based on the Johnson factors, to trustee compensation under § 330(a). See, e.g., Garb v. Marshall (In re Narragansett Clothing Co.), 210 B.R. 493, 497 (1st Cir.BAP1997); In re Draina, 191 B.R. 646, 649 (Bankr.D.Md.1995). We therefore apply the adjusted lodestar test here.
At the outset, we note Harris Trust’s failure to cross-appeal from either the hourly rate determined by the district court, or the multiplier it applied. Given the fact that Harris trust did not cross-appeal, our review is limited only to determining whether the record justifies the increase in either the hourly rate or the multiplier contended for by Connolly. We have no occasion to consider whether a reduction might be required on this record.
a. Hourly rate and hours expended
Judicial review of the bankruptcy court’s factual determinations in connection with a fee award is highly deferential:
When we scrutinize factual determinations and discretionary determinations made by a bankruptcy judge, such as may be involved in calculating and fashioning appropriate fee awards, we give considerable deference to the bankruptcy court[.] Historically, bankruptcy courts have been accorded wide discretion in connection with fact-intensive matters, and in regard to the terms and conditions of the engagement of professionals .... The bankruptcy judge is on the front line, in the best position to gauge the ongoing interplay of factors and to make the delicate judgment calls which such a decision entails.
Casco N. Bank, N.A. v. DN Assoc. (In re DN Assoc.), 3 F.3d 512, 515 (1st Cir.1993) (quotation omitted).
The bankruptcy court initially determined that a uniform, reasonable hourly rate of $500 should be applied to the 1,779 hours reasonably expended by Connolly, resulting in a basic lodestar fee of $889,500. The district court rejected the $500 per hour rate as unsupported, concluding that the hourly rates used should vary according to the work done. When upon remand the bankruptcy court adjusted the hourly rate to a uniform $400 per hour, however, the district court abandoned the variable rate concept and approved the lower, uniform rate.
A Chapter 7 trustee does perform a variety of functions in his role, including investigating, liquidating, and distributing estate assets. In re Colburn, 231 B.R. 778, 783 (Bankr.D.Ore.1999) (citing 11 U.S.C. § 704). Where the trustee has performed work that differs in complexity, a solution is to adjust his fee as a whole, to arrive at a “blended” rate. Garb, 210 B.R. at 499. We agree that the appropriate approach here was a unified rate for all of the trustee’s services.
The record, however, does not justify the rate of $500 per hour that Connolly seeks. This is true for several reasons. First, the bankruptcy court rejected expert testimony Connolly presented to justify a rate of $500 per hour. Second, it found that the top rate generally charged by private attorneys for bankruptcy work in the area was $350 per hour. MiniS-[1245]*1245cribe, 257 B.R. at 62. The bankruptcy court acknowledged the absence of specific evidence from which a rate of $500 per hour could be justified. Given that Harris Trust has not cross-appealed the district court’s determination that $400 per hour was a reasonable rate, however, we will affirm that determination as unchallenged.
b. Lodestar multiplier
The bankruptcy court further concluded, in light of the exceptional results obtained in this case and other Johnson factors, that a multiplier was required to adequately compensate Connolly for his efforts. The district court disapproved of a 3.5 multiplier, however, concluding that the bankruptcy court had overstated the risk of non-payment run by Connolly. We also conclude that the 3.5 multiplier was excessive. The 3.5 multiplier risks double payment or overpayment for the services provided. Cf., e.g., Rosenbaum v. MacAllister, 64 F.3d 1439, 1447-48 (10th Cir.1995) (stating “our conscience is shocked by an award of a 3.16 multiplier that results in a fee equal to more than $900 per hour for every attorney, paralegal, and law clerk who worked on the case”).
In its second decision, the bankruptcy court lowered the multiplier to 2.57, relying on the multiplier used for the attorney fee award in Gottlieb v. Wiles, 150 F.R.D. 174 (D.Colo.1993), subsequently reversed for application of a common fund award in Gottlieb v. Barry, 43 F.3d 474 (10th Cir.1994). The 2.57 multiplier, unchallenged by Harris Trust, finds some support in other lodestar multiplier cases. See, e.g., In re Computron Software, Inc., 6 F.Supp.2d 313 (D.N.J.1998) (applying, in securities fraud class action, lodestar test as cross check for common fund recovery and finding 2.5 multiplier “fair”); In re Biskup, 236 B.R. 332, 337 (Bankr.W.D.Pa.1999) (applying fee that worked out to 2.76 multiplier for exceptional work by trustee in discovery and preservation of asset in bankruptcy case). Given the strong findings by the bankruptcy court in support of its determination that the results achieved were sufficiently extraordinary to justify a multiplier, and the reasoning of the district court flowing from these findings, we cannot conclude, based on our standard of review, that the district court erred by applying a 2.57 lodestar multiplier.
Harris Trust contends that Connolly improperly awarded himself the full $3,044,953.69 nearly three years ago, and should be compelled to disgorge any excess amounts, plus interest. These are matters that can be determined in further proceedings. The judgment of the United States District Court for the District of Colorado is AFFIRMED. Appellant’s motion to file a supplemental appendix is granted; as is appellee’s motion for leave to file a surreply.