B.D. PARKER, JR., Circuit Judge.

introduction

This appeal arises from a decision of the United States District Court for the Southern District of New York (Stanton, /.), acting under the Broadcast Music, Inc. (“BMI”) Consent Decree, 1 to set a rate for Music Choice’s licensing of BMI’s music. The license would apply to BMI music used by Music Choice on its cable, satellite, and Internet services between October 1,1994 and September 30, 2004. Since BMI and Music Choice were unable to agree on a rate, the Consent Decree required the court to set one.

The District Court entered its first decision setting a rate in 2001. See United States v. Broad. Music, Inc., No. 64 Civ. 3787(LLS), 2001 WL 829874, 2001 U.S. Dist. LEXIS 10368 (S.D.N.Y. July 23, 2001) (“Music Choice I”). In that decision, the District Court rejected BMI’s proposed blanket license fee of 3.75%, and fixed the rate at 1.75%, less than half the rate established in a deal between BMI and DMX, a competitor of Music Choice. The District Court reasoned that the price *93 paid for music by retail customers that was the basis for the rate set under BMI’s agreement with DMX did not reflect the fair market value of the music to the extent that price included both the cost of the music itself as well as the cost of actually delivering the music to retail customers. The Court concluded that the fair market value of the music was better expressed by the wholesale price at which Music Choice sold to cable and satellite operators. Id. 2001 WL 829874 at *6-7, 2001 U.S. Dist. LEXIS 10368 at *21-23. On appeal, we vacated and remanded the decision to permit the District Court to reassess its calculation of the fair market value of the disputed music rights. See United States v. Broad. Music, Inc., 316 F.3d 189 (2d Cir.2003) (“Music Choice II”). On remand, the District Court set the rate incorporating retail value as a component of the value of the music rights. See United States v. Broad. Music, Inc., No. 64 Civ. 3787(LLS), 2004 WL 1171249, 2004 U.S. Dist. LEXIS 9461 (S.D.N.Y. May 26, 2004) (“Music Choice III ”). This appeal followed. Because we believe that the District Court misinterpreted the scope of our previous opinion, we again remand to permit the District Court to exercise its unconstrained reconsideration.

background

The history of this rate dispute is set out in comprehensive detail in Music Choice I, Music Choice II, and Music Choice III. Familiarity with these decisions is presumed. Music Choice transmits 55 different music channels, commercial-free, to listeners’ televisions via cable and satellite, and to their computers via the Internet. Music Choice is a partnership between wholly-owned subsidiaries of Adelphia Cable, Comcast Cable, Cox Cable, EMI Group, Motorola Broadband Communications Sector, Microsoft Corporation, Sony Corporation of America, and Time Warner, Inc.

BMI is one of the two major performing rights societies which license the public performing rights to most copyrighted musical works in this country. ASCAP, its chief competitor, is the other. 2 BMI typically issues blanket licenses to broadcast any and all of the approximately 4.5 million musical works in its portfolio for a finite period of time. Because of the inherently anti-competitive conditions under which BMI and ASCAP operate, they are regulated by court-approved consent decrees. See BMI Consent Decree; ASCAP Consent Decree. 3 In 1994, the BMI consent decree was modified to create a rate court mechanism to fix reasonable license fees in the event BMI and its customers were unable to do so. From October 1994 through January 1997, Music Choice had an interim license agreement with BMI that was rolled over from month to month. However, when Music Choice applied to BMI for a blanket license for cable, satellite and Internet distribution for the ten-year period of October 1994 through September 2004, the parties were unable to agree on terms and resorted to the rate court.

*94 A rate court’s determination of the fair market value of the music is often facilitated by the use of benchmarks — agreements reached after arms’ length negotiation between other similar parties in the industry. See Music Choice II at 194. A good deal of the previous litigation in this case has focused on which agreements could serve as appropriate benchmarks for Music Choice and BMI. The main candidates were: (1) a 1990 Licensing Agreement Between BMI and Music Choice, (2) a 1995 Licensing Agreement Between BMI and DMX (“the DMX Agreement”), (3) a 2002 Licensing Agreement Between Music Choice and ASCAP, and (4) the rate BMI charged its radio and Internet licensors. 4

In proceedings in the District Court leading to Music Choice I, BMI argued that the court should use the 1995 DMX Agreement as a benchmark and set the rate at 3.75% of Music Choice’s gross revenues. Music Choice argued unsuccessfully for a lower rate based on what BMI charged radio broadcasters and Internet licensees. Id. 2001 WL 829874 at *2, 2004 U.S. Dist. LEXIS 9461 at *7. The disagreement between BMI and Music Choice centered on the question of whether the DMX Agreement provided an appropriate benchmark for Music Choice’s rate. Music Choice argued that the DMX Agreement resulted from circumstances specific to Music Choice’s competitor DMX and did not reflect the relative bargaining positions of Music Choice and BMI. Although BMI and DMX originally had an agreement that was essentially identical to the 1990 agreement between BMI and Music Choice (2.0-2.1% of wholesale revenues plus 2.0-2.1% of the cable operators’ gross revenues), sometime before the expiration of that agreement, DMX and BMI disagreed whether DMX was obligated to count the cost of hardware sold to retail customers towards the revenues of the retailers (2.0-2.1% of which it was obligated to pay to BMI). In Music Choice I, the District Court found that DMX’s strained financial situation made it eager to arrive at a deal with BMI, even if disadvantageous, so long as DMX was guaranteed that it would not pay more than its competition. DMX ultimately agreed to pay BMI half the amount BMI claimed it was owed in the hardware dispute, and the parties entered into a new license agreement (the DMX Agreement) which required that DMX pay BMI 3.75-4.00% of gross revenues.

The District Court rejected the DMX Agreement’s rate as a basis for setting the rate for Music Choice. The court reasoned that the DMX Agreement was based on the retail price of the music and did not reflect the fair market value of the music to the extent that price included both the cost of the music itself as well as the cost of actually delivering the music to retail customers. It concluded that the fair market value of the music was better expressed by the wholesale price at which Music Choice sold to cable and satellite operators. Id. 2001 WL 829874 at *7-8, 2001 U.S. Dist. LEXIS 10368 at *24-25. The District Court held that “the concept on which the 1995 DMX rate agreement rests- — that the license fees should capture *95 a portion of the cable operators’ revenues — is flawed, and should be disregarded in considering that agreement as a reference point.” Id. 2001 WL 829874 at *7, 2001 U.S. Dist. LEXIS 10368 at *23. The District Court appeared to set its final rate, 1.75% of Music Choice’s gross revenues, because that rate reflected DMX’s rate once the cable operators’ revenues were removed, id. 2001 WL 829874 at* 7, 2001 U.S. Dist. LEXIS 10368 at *23, and was the same rate that BMI charged its Internet licensees, id. 2001 WL 829874 at *7-8, 2001 U.S. Dist. LEXIS 10368 at *24-25.

On appeal, in Music Choice II, we remanded because we concluded that the District Court should not have rejected the retail price of music as an indication of its fair market value. We said the District Court erred by finding “that retail price was not a good indicator of fair market value because the retail seller incurred, and needed to cover, the costs of processes and services necessary to bring the music to market, which were not provided by the copyright holders.” Music Choice II at 195. Other than that, we “expressed] no view [as to] what would be a proper rate.” Id. at 197.

On remand and after further fact-finding, the District Court read Music Choice II to endorse the DMX Agreement as reflecting “normal competitive market terms.” Music Choice III, 2004 WL 1171249 at *1, 2001 U.S. Dist. LEXIS 10368 at *4. It then reversed its previous course and adopted the rate of 3.75% of Music Choice’s gross revenues set in the 1995 DMX Agreement. It reasoned that “the 3.75% range is not novel, but well established in the industry agreements and practices.” Id. 2004 WL 1171249 at *2, 2001 U.S. Dist. LEXIS 10368 at *5. However, the District Court reached this contrary conclusion without explaining why its earlier rejection of that rate, including its previous concern about the market conditions surrounding the DMX Agreement, was wrong. This appeal followed.

DISCUSSION

7. Standard of Review

The rate court is responsible for establishing the fair market value of the music rights, in other words, “ ‘the price that a willing buyer and a willing seller would agree to in an arm’s length transaction.’ ” Music Choice II at 194 (quoting ASCAP v. Showtime / The Movie Channel, Inc., 912 F.2d 563, 569 (2d Cir.1990) (“Showtime ”)). In choosing a benchmark and determining how it should be adjusted, a rate court must determine “the degree of comparability of the negotiating parties to the parties contending in the rate proceeding, the comparability of the rights in question, and the similarity of the economic circumstances affecting the earlier negotiators and the current litigants,” United States v. ASCAP (Application of Buffalo Broad. Co., Inc.), No. 13-95(WCC), 1993 WL 60687 at 18, 1993 U.S. Dist. LEXIS 2566, at *61 (S.D.N.Y. Mar. 1, 1993), as well as the “degree to which the assertedly analogous market under examination reflects an adequate degree of competition to justify reliance on agreements that it has spawned.” Showtime, 912 F.2d at 577. Although, under the terms of the BMI Consent Decree, BMI bears the burden of establishing the reasonableness of its rates, the setting of appropriate rates remains the responsibility of the District Court. 5

*96 We review the rate set by the District Court for reasonableness. Music Choice II at 194; Showtime, 912 F.2d at 569; Broadcast Music, Inc. v. Columbia Broad. Sys., Inc., 441 U.S. 1, 24, 99 S.Ct. 1551, 60 L.Ed.2d 1 (1979) (holding generally that blanket and per-program licenses by ASCAP and BMI were not per se antitrust violations, but rather these licenses, “when attacked, ... should be subjected to a moi"e discriminating examination under the rule of reason”). Our review is a two-part task. In order to find that the rate set by the District Court is reasonable, we must find both that the rate is substantively reasonable (that it is not based on any clearly erroneous findings of fact) and that it is procedurally reasonable (that the setting of the rate, including the choice and adjustment of a benchmark, is not based on legal errors). As we explained in both Music Choice II and Showtime, this substantive and procedural review:

is a factual matter, albeit a hypothetical one, but at the same time that the factual component of rate setting does not render all aspects of the district court’s decision subject to review under the ‘clearly erroneous’ standard. This is because in making a factual determination, a decision-maker might rely on legally impermissible factors, fail to give consideration to legally relevant factors, apply incorrect legal standards, or misapply correct legal standards.

Music Choice II at 194-195 (citations and internal quotation marks omitted).

As we held with respect to ASCAP, rate-setting courts must take seriously the fact that they exist as a result of monopolists exercising disproportionate power over the market for music rights. 6 Thus, we review the District Court’s evaluation of the facts surrounding the formation of the benchmark agreements, including the credibility of witnesses and other evidence at trial, for clear error. The adjustment of the benchmark to best approximate the fair market value of the music may be based on factual findings, but also may contain legal conclusions that we review de novo.

II. Music Choice II

The District Court read our decision in Music Choice II to hold that the DMX Agreement “reflected normal competitive market terms better than [the court] allowed [in Music Choice I ].” Music Choice III, 2004 WL 1171249 at *1, 2004 U.S. Dist. LEXIS 9461 at *4. Because we held that the retail value of the music was an appropriate component of the fair market value, the District Court appears to have concluded that we were endorsing the DMX Agreement because it included a retail component. Id. As a result of this reading of Music Choice II, the District Court added the retail value of the music in the DMX Agreement to its original rate, and set the new rate at 3.75% of Music Choice’s gross revenue. It reasoned that, as a result of Music Choice II, “the 2% [the court] deducted from the rate, representing the portion of the BMI-DMX rate which derived from retail consumers, must be restored, because ‘retail revenues derived from the sale of the music fairly measure the value of the music.’ ” Id. 2004 WL 1171249 at 1, 2004 U.S. Dist. *97 LEXIS 9461 at *3 (quoting Music Choice II at 195).

These conclusions constitute an over-reading of our decision. The District Court had previously said of the DMX Agreement: “DMX had no palatable licensing alternatives to accepting a blanket license from BMI.” Music Choice I 2001 WL 829874 at *5, 2004 U.S. Dist. LEXIS 9461 at *19. We could not overturn the District Court’s previous finding of facts surrounding the formation of the DMX Agreement without establishing that they were clearly erroneous, something we did not do. See Fed. R. Civ. Pro. 52(a). We confined our discussion to the role of retail value in establishing the fair market value of the music rights, and even that holding permitted the District Court to approximate retail value if it found certain facts on remand. Our essential holding in Music Choice II was that the District Court was incorrect in finding “that retail price was not a good indicator of fair market value because the retail seller incurred, and needed to cover, the costs of processes and services necessary to bring the music to market which were not provided by the copyright holders.” Music Choice II at 195. However, we explicitly refrained from expressing a view on what the appropriate rate should be. Id. at 197. Further, in spite of our endorsement of retail price as generally a good marker for fair market value, we did not require it to be used in all circumstances, but only “absent some valid reason for using a different measure.” Id. at 195. We based our holding on the facts developed to that point. However, given that the District Court conducted additional fact finding on remand, it was free to find fair market value on any basis adequately supported by the record, including, of course, any new facts developed.

For example, in Music Choice II we held that it would be appropriate for the District Court to “approximate fair market value on the basis of something other than the prices paid by consumers” if, “where the customers pay a single fee for a package of audio and visual programming, which includes the music,” the District Court could not “determine what part of the fees paid was for the music, as opposed to other programming.” Id. at 195 n. 2. Additionally, we said that “if it were demonstrated that retail purchasers were motivated to pay more because of advantages that resulted from a particular mode of delivery, such as better quality, better accessibility or whatever, this might justify a conclusion that retail price of the service purchased by the customer exceeded the fair market value of the music.” Id. at 196 n. 3. We also observed that an approximation based on wholesale revenue might be appropriate in a case such as this, “where retail revenues attributable to the music are difficult to ascertain because of the bundled packages offered at retail, while wholesale revenues attributable to the music are easily determined” and noted that this “may be a useful way to proceed.” Id. at 197 n. 5. These observations demonstrate that, while we emphasized the importance of retail price in determining the fair market value, we did not require that this retail price be drawn from any particular agreement if the District Court found facts that made retail value difficult to isolate or inferior to other indices. This flexibility available to the District Court is particularly important in light of the District Court’s adoption in Music Choice III of the DMX Agreement’s rate despite its previous misgivings about that Agreement.

It is important to note that in Music Choice II we did not make the DMX Agreement a paradigm. Instead, we acknowledged the District Court’s own concerns, expressed in Music Choice I, about *98 the soundness of the DMX Agreement as a benchmark, recalling that “DMX’s strained financial situation made it especially eager to arrive at a deal with BMI, even if disadvantageous, so long as DMX was guaranteed (by a most favored nation provision) to be no worse off than its competition.” Id. at 193. We also noted that “[t]he parties contest whether the BMI-DMX deal provides an appropriate benchmark for Music Choice’s rate, or whether idiosyncratic circumstances distorted that negotiation so that the resulting agreement does not accurately reflect the relative bargaining strengths of Music Choice and BMI.” Id. at 192.

We are particularly concerned that the District Court’s view of our holding may have led it to draw conclusions in Music Choice III that were either incongruent with factual findings critical to Music Choice I or not fully supplemented by whatever additional facts the District Court may have found on remand. These problems interfere with our evaluation of the substantive reasonableness of the rate set. We are conscious that rate court decisions have effects that reach beyond the litigants involved in the proceedings. Such decisions directly affect all participants in the industry since the Consent Decree obligates BMI to offer rates paid by one user to similarly situated users. See BMI Consent Decree, art. XIV(C). For these reasons, our concern with the reasonableness of rates set is heightened. We therefore conclude, without expressing any opinion on the substantive reasonableness of the rate set by the District Court, that the process of ascertaining the rate was flawed due to a misperception of our holding in Music Choice II. Consequently, we remand for additional consideration.

III. Remand

On remand, in revisiting the rate it set, if the District Court again determines that the DMX Agreement is an appropriate benchmark, it may wish to consider the market conditions at the time the parties were negotiating 7 and any particular features of the business models of Music Choice or DMX that make them more or less similar. The District Court might also wish to consider whether the DMX Agreement’s reflection of the retail value of the music in that transaction also reflects the value of the music rights licensed by Music Choice — in other words, whether the DMX Agreement adequately took into account the difficulty in isolating the retail value of the music rights themselves, which were bundled with other rights when sold.

Additionally, the District Court might choose to clarify whether it still rejects BMI’s contention that Music Choice is unique in its “intensity of use” of the music rights it licensed from BMI. In Music Choice I, the District Court dismissed BMI’s argument that Music Choice must pay more because of its “intensity of use” of the music rights, reasoning that the “distinction cannot be made with respect to Internet music services, a number of which have the same characteristics.” Music Choice I, 2001 WL 829874 at *8 n. 20, 2001 U.S. Dist. LEXIS 10368 *25 n. 20. The role of BMI’s Internet licensees in setting a licensing rate for Music Choice was not definitively determined by our holding in Music Choice II. 8 We emphasize *99 that, on remand, the District Court is free to fix a rate by reference and adjustments to any benchmark it deems appropriate. 9 It need only explain how it reached a particular rate sufficiently to permit our review of the rate for reasonableness, should we be required to do so.

conclusion

We vacate and remand to the District Court for further proceedings consistent with this opinion.

Notes

. See Amended Final Judgment entered in United States v. Broadcast Music, Inc., 1966 U.S. Dist. LEXIS 10449, 1966 Trade Cas.(CCH) P71,941 (S.D.N.Y.1966), modified by 1994 WL 901652, 1994 U.S. Dist. LEXIS 21476, 1996-1 Trade Cas. (CCH) P71,378 (S.D.N.Y.1994) (the "BMI Consent Decree”). The BMI Consent Decree requires BMI to make through-to-the-listener licenses available for public performances of its music and to provide applicants with proposed license fees upon request. If BMI and the applicant cannot agree on a fee, either party may apply to the rate court for the determination of a reasonable fee.

. As discussed in detail in the previous Music Choice decisions, individual copyright owners assign BMI and the American Society of Composers, Authors, and Publishers ("AS-CAP”) the right to license nondramatic public performances of their musical compositions. Performing rights societies in turn issue blanket licenses entitling licensees to perform any and all works in the societies' repertories, for a finite period of time. See, e.g., Music Choice lint 190.

. United States v. ASCAP, 1940-43 Trade Cas. para. 56,104 (S.D.N.Y.1941), as amended, United States v. ASCAP, 1950 WL 42273, 1950-51 Trade Cas. para. 62,595 (S.D.N.Y.1950).

. The 1990 Licensing Agreement Between BMI and Music Choice required Music Choice to pay a license fee equal to 2% of its gross revenue plus 2% of the cable operators' gross revenues from Music Choice's service (minus the operators' payment to Music Choice) for the first two years; both percentages increased to 2.1% for the third year. The relevant rate from the DMX Agreement was 3.75%. The 2002 Licensing Agreement Between Music Choice and ASCAP, which was made after the District Court’s decision in Music Choice I, required Music Choice to pay 1.75% of Music Choice’s revenue. BMI charged its radio and internet licensors a rate equal to or less than 1.75%.

. The BMI consent decree stipulates that "[i]n any [rate] proceeding, [BMI] shall have the burden of proof to establish the reasonableness of the fee requested by it. Should [BMI] not establish that the fee requested by it is a reasonable one, then the Court shall deter *96 mine a reasonable fee based upon all the evidence.” BMI Consent Decree, art. XIV(A).

. “Though the rate court's existence does not mean that ASCAP has violated the antitrust law, the court need not conduct itself without regard to the context in which it was created.... The disinfectant [of the rate court] need not be a placebo.” Showtime, 912 F.2d at 570.

. The District Court might, for example, find facts to suggest that our decision on "carve-out” licenses changed market conditions and adjust a benchmark on that basis. See United States v. Broad. Music, Inc. (Application of Muzak LLC, AEI Music Network), 275 F.3d 168, 171 (2d Cir.2001).

. In Music Choice II, we expressed skepticism about the utility of using the rate that BMI *99 charged Internet licensees as a benchmark, but did not reverse the District Court’s earlier findings about the nature of use of music rights by Music Choice. We did, however, emphasize the difference in scope, observing that Internet licensees reached far fewer customers than Music Choice’s services. We discussed the rate BMI charged its Internet licensees in a brief aside and noted that ”[w]e do not mean to suggest, however, that the Internet rate [charged by BMI for music distribution over the Internet] supplies a fair benchmark” and observed that "Internet distribution is in its infancy.” Id. at 197 n. 6. Even in light of these observations, as long as the District Court includes retail value of the music in its valuation of the music rights (either as enshrined in a previous Agreement or as an approximation of the wholesale price or any other way it finds reasonable), the District Court is free to look to the Internet licensees and make explicit its findings concerning the role that the "intensity of use,” by those licensees and by Music Choice, plays in its rate setting.

. Appellant argues that the District Court erred in not using the ASCAP 1.75% rate as a benchmark. The District Court rejected the ASCAP rate, reasoning that "[i]t is apparent that the 2002 license agreement between Music Choice and ASCAP rested so heavily upon the rate set in my 2001 decision, later vacated by the Court of Appeals, that it cannot be used as a valid benchmark for purposes of the present proceeding." Music Choice III, 2004 WL 1171249 at *3, 2004 U.S. Dist. LEXIS 9461 at *3. We find no error in this part of the District Court’s holding.