This, appeal arises from the decade-long effort to expand the federal government’s facilities for research into animal diseases at Plum Island, a small, federally owned island just off the eastern tip of Long Island’s north fork. The appellants, Seaboard Surety Co. and The Home Insurance *959Co. (“the sureties”), provided a bond pursuant to the Miller Act, 40 U.S.C. § 270a et seq., guaranteeing the performance of The Joseph Morton Company (“Morton”), which was the general construction contractor on that project. The sureties were held liable based on Morton’s breach of its contract with the United States for the Plum Island expansion. On appeal they challenge the jurisdiction of the district court, the admission of certain evidence of bad faith on the part of the sureties, and the calculation of the prejudgment interest awarded by the trial court.
BACKGROUND
In the early 1970s, the Department of Agriculture (“DOA”), which operates the government’s Plum Island Animal Disease Center, began making plans to expand that facility. The first step was to develop criteria to be used in the design and eventual construction of the new buildings. Central to those criteria, and to subsequent events, was the goal of insuring that Plum Island be operated as a “bio-containment” facility; that is, that the extremely dangerous strains of animal diseases being studied not become hazardous to the surrounding environment. It was thus necessary to design air-tight buildings, using sophisticated ventilation and filters.
Once the criteria had been established, the government in 1973 hired Perkins & Will, an architectural and engineering firm, to prepare plans and specifications for the proposed facilities which would meet the design criteria. The Perkins & Will designs were accepted by the government and became the basis -for an invitation for bids issued by the DOA on August 4, 1976. The low bid was submitted by Morton, which on September 24, 1976, was awarded the contract for $10,049,000, subsequently increased to $10,689,000.
Construction on the project began in March 1977, and was almost immediately beset by problems. On March 23,1977, the government issued the first in what quickly became a series of “cure notices”, which put a contractor on notice that some aspect of its performance is unsatisfactory and which require certain specified corrective action to be taken by a fixed date.
Throughout the time Morton was on the job, there were various problems that culminated on February 14, 1979, with the issuance of a final, 55-page cure notice, detailing many inadequacies in Morton’s work, including defective construction, poor maintenance, and undue delays. Unsatisfied with Morton’s response, the government terminated Morton for default on March 1, 1979.
At the time of termination, the government had approved payment to Morton, on an as-completed basis, of $9,048,060, of which it had actually paid $8,231,260. The sureties claim that approximately 85% of the total project was completed, since the amount approved for payment represented 85% of the total contract price. As the government notes, however, none of the work on such a project is officially accepted or approved until the entire project '.s complete.
In any event, the government next turned to the sureties and requested that they step in and complete the project. After offering to complete the project using Morton, an option the government understandably rejected, the sureties, in letters dated April 13, 1979 and May 2, 1979, declined to complete. As we will discuss more fully infra, a surety has the option, when its principal is default-terminated, of either coming in and completing the project or paying the government its damages, essentially the cost of completion. Trinity Universal Insurance Company v. United States, 382 F.2d 317, 321 (5th Cir.1967), cert. denied, 390 U.S. 906, 88 S.Ct. 820, 19 L.Ed.2d 873 (1968). A surety may, of course, also challenge the propriety of the default termination, thereby, in effect, denying liability on the bond.
This latter course was followed by the sureties in this case. On May 30, 1980, Seaboard sent to Mr. Calvin Persinger, the government’s contracting officer on the project, a letter stating, “Seaboard Surety Company ... shares the position of its principal ... that the default determination *960was unwarranted and that the contract was not breached by Morton, but by the Government. Accordingly, Seaboard denies any liability under its performance bond.” Apparently, Seaboard’s denial of liability spoke for its co-surety, The Home Insurance Co., and there is no issue raised as to any difference in the positions of the two sureties for purposes of this appeal.
After the sureties declined either to complete the project or pay the cost of completion, the government hired Lasker-Goldman Corporation, a joint venture of LaskerGoldman Company and Goldman Associates, as “construction manager” to oversee the anticipated completion and assess the cost and magnitude of completing the work. As a result of Lasker-Goldman’s assessment and a shortfall in funds available for completion of the project, it was eventually decided that two of the buildings covered by the original contract, a diagnostic research laboratory (DRL) and a vaccine research laboratory (VRL), would not be completed.
In December 1980, congress appropriated an additional $10,100,000 for completion of the project, and by 1985, it had been finished, except for the VRL and DRL. The cost of completion was almost as high as the initial cost of the project, despite the fact that 85% of the original appropriation had been paid to Morton, and despite the fact that two of the five buildings planned were never completed. Thé government attributes the high cost of completion to inflation, poor work by Morton that had to be corrected if the facility was adequately to contain the viruses being studied, and deterioration of the material left on the site, where it sat, apparently unattended, from March 1979, when Morton was terminated, to sometime in 1980, when LaskerGoldman took over. The sureties claim that Lasker-Goldman was vastly overpaid (over $4,600,000) for its role as construction manager. In any case, the DOA ultimately spent in the neighborhood of $20 million to build 60 percent of a project initially budgeted at $10,689,000.
As might be expected, these events gave rise to multiple legal proceedings. In March 1980 Morton filed for bankruptcy under Chapter 11, a petition that was dismissed in April 1983 when the bankruptcy court refused to discharge Morton. In 1980 Morton, its principal Joseph J. Battaglia, and a Morton employee were indicted in connection with the submission of fraudulent documents to the DOA as part of a change order Morton submitted on the Plum Island project. In April 1981 they were convicted of conspiracy to defraud the United States, in violation of 18 U.S.C. § 371, making false statements in the change order in violation of 18 U.S.C. § 1001, and obstructing justice, 18 U.S.C. § 1503. These convictions were upheld on appeal. United States v. Joseph Morton Company, et al., No. CR-80-413 (E.D.N.Y. April 24, 1981), aff'd, Nos. 81-1174, -1233 (2d Cir. Jan. 11, 1982).
Morton took the offensive on March 1, 1982, filing a complaint in the court of claims (now the claims court), seeking determinations that it had not breached its contract with the United States and that the termination was one “for the convenience of the government.” On the same day, the government filed the instant suit against the sureties, seeking to enforce their obligations based on Morton’s breach.
In July 1983, while proceedings in the instant suit were stayed, the claims court ruled that the criminal conviction of Morton, even though it came after the termination, justified the government’s conduct as a default termination. Joseph Morton Co. v. United States, 3 Cl.Ct. 120 (1983). This holding was affirmed by the federal circuit, Joseph Morton Company, Inc. v. United States, 757 F.2d 1273 (Fed.Cir.1985). Neither the claims court nor the federal circuit reached the issue of whether termination was justified based on Morton’s poor performance. Id. at 1275.
Following the resolution of Morton’s claims court action, proceedings went forward in the instant case. After extensive discovery, a jury trial began on May 7, 1986. The major focus of the case was Morton’s performance from 1977 to 1979, with the government seeking to show that Morton had flagrantly breached the con*961tract, doing poor work, doing even thát late, failing to pay subcontractors, and misleading the government about the work it had done. In addition, the government sought to show bad faith on the part of the sureties in performing their obligations.
For their part, the sureties sought to establish that while Morton’s work was far from perfect, termination due to deficient performance had not been justified. They also attempted to show that the original Perkins & Will design had been fundamentally flawed, rendering performance by Morton (in constructing a functioning bio-containment facility) impossible, and that the completion costs incurred by the government were unreasonable.
After two weeks of trial, the jury returned a verdict for the government of $10,631,326. To this sum. the district judge added $6,755,321.73 in prejudgment interest, calculated from May 30, 1980, the date Seaboard sent Mr. Persinger the letter denying liability under the bond, and entered judgment for $17,386,647.73. The sureties appeal.
DISCUSSION
A. Jurisdiction of the District Court.
The first contention raised by the sureties is that the district court lacked subject matter jurisdiction because claims against a surety are limited by the Contract Disputes Act (“CDA”), 41 U.S.C. § 601 et seq., to the claims court. This argument takes two forms. First, the sureties urge that a surety is a “contractor” within the meaning of the CDA, and thus that the act, by its terms, is applicable to a surety’s performance bond. Second, they argue that even if a surety is not a “contractor”, the government must press its claim against the sureties in the claims court, because a surety’s liability is wholly derivative from its principal’s liability, and Morton, the principal, unquestionably was the CDA contractor. These arguments are unconvincing.
1. Applicability of the CDA to a Miller Act Surety.
We have little difficulty rejecting the first of the sureties’ jurisdictional arguments, for it is clear from the language of the CDA and from its legislative history that a surety is not a “contractor” for CDA purposes.
Section 3 of the CDA, 41 U.S.C. § 602, defines the contracts covered by the act. They include contracts for:
(1) the procurement of property, other than real property in being;
(2) the procurement of services;
(3) the procurement of construction, alteration, repair or maintenance of real property; or,
(4) the disposal of personal property.
A Miller Act performance bond fits none of these categories. The only one that arguably might include a surety is (2), a contract for the procurement of "services.” But even if the surety agreement could be considered a contract to procure “services”, it is not a contract to which the government is a party. The bond, itself, was signed by the sureties and Morton; the government’s rights are as a third-party beneficiary, not as a party. It is true that the Miller Act required Morton to obtain the performance bond, 40 U.S.C. § 270a(a)(1), but that fact alone does not make the government a party to the bond. Since the CDA requires that the “contract” sued upon be a government contract, see 41 U.S.C. §§ 601(4), 602(a), we conclude that it does not apply to a bond issued by a surety to a contractor. See also Universal Surety Company v. United States, 10 Cl.Ct. 794, 800 (1986) (surety is “not a ‘contractor’ under CDA”).
Nothing in the legislative history alters this conclusion. The stated purpose of the CDA was to make uniform the system of dispute resolution governing federal contracts, a system that varied from agency to agency and from contract to contract. Congress felt that this system discouraged businesses from bidding on government contracts, and that its inefficiency almost certainly increased the cost of government projects. S.Rep. No. 95-1118, 95th Cong., 2d Sess., reprinted in 1978 U.S.Code Cong. & Ad.News 5235, 5236-38. Nowhere are sureties or the Miller Act mentioned, and *962the goals of encouraging bids on contracts and lowering the cost of resolving disputes surrounding government contracts bear little relation to the type of performance bonds that are issued for all large-scale government construction projects. Congress simply had an entirely different set of problems in mind when it passed the CDA, and we see no reason to judicially transform sureties into “contractors” where congress has not done so.
2. “Derivative Liability ” of a Surety.
A closer question is presented by the sureties’ argument that the government should be forced to proceed in the claims court because any liability on the part of the sureties is derivative from Morton’s liability and Morton is undeniably a “contractor” under the CDA. Many of the premises that support the argument, looked at in isolation, are valid. Taken together, however, these premises do not lead to the conclusion urged by the sureties.
It is certainly correct that Morton was a “contractor” within the meaning of the CDA. It is also true that a surety’s liability is derivative from its principal’s liability. Coleman Capital Corp. v. Travelers Indemnity Co., 443 F.2d 47, 52 (2d Cir.1971); Atterbury v. Carpenter, 321 F.2d 921, 923-24 (9th Cir.1963).
. It does not follow, however, that the derivative nature of a surety’s liability places it in its principal’s shoes for purposes of vesting exclusive jurisdiction in the claims court. The sureties’ liability is not the same as Morton’s; it is merely secondary to Morton’s. For example, it would easily be possible for Morton to be in breach in a manner that would not leave the sureties liable. Indeed, the federal circuit found Morton in breach based solely on the fraud conviction, see Morton, 757 F.2d at 1279, forcing the government at the trial below to show a performance-based breach to establish the sureties’ liability.
Since the liability of a surety is distinct from that of its principal, there is no compelling reason to conclude that a claim by the government against a surety must be tried in the claims court simply because claims against the principal must be heard there under the terms of the CDA.
The sureties argue further that any trial of their liability will necessarily focus primarily on the performance of a contractor, as did the trial below. They contend that congress intended the adjudication of such issues to be centralized in a single forum, which would soon develop special expertise in the handling of government contracts cases.
While there is some support in the legislative history for the notion that congress wished to consolidate cases involving claims against and by government contractors, the underlying purpose served by the legislation was to protect contractors, not by creating a court of special competence, but by standardizing the procedures to be followed when contractors and the government have a dispute. The principal rationale for vesting jurisdiction in the claims court thus vanishes where the claim is against a surety rather than a contractor.
Moreover, while the issues in cases like the instant suit will center on the performance of the contractor, there is no reason to believe that the district courts are not fully capable of trying such claims. When it divested the district courts of jurisdiction over cases involving contractors, congress was concerned not with the competence of the district courts in trying such cases, but rather with improving the efficiency of the system as it actually affected contractors. We have every confidence that the district courts can handle even the complexities of government contract and suretyship law, as they have consistently done in the past. See, e.g., United States v. Seaboard Surety Co., 339 F.2d 1 (2d Cir.1964); United States v. Glens Falls Ins. Co., 546 F.Supp. 643 (N.D.N.Y.1982); United States v. Warren Corp., 624 F.Supp. 1163 (D.Mass.) vacated, 805 F.2d 449 (1st Cir.1986).
In short, the CDA vests jurisdiction in the claims court over claims arising from government contracts when such claims involve, as parties, the contractors themselves. The goal was to protect contrac*963tors and make more smooth the process for resolving their disputes with the government. However, since sureties are not contractors under the CDA, and since liability on the part of a surety is distinct from that of its contractor, we conclude that the district court properly exercised jurisdiction in this case.
B. Admissibility of Evidence of the Sureties’ “Bad Faith”.
The sureties next argue that it was reversible error for the district court to admit evidence of “bad faith” on their part in denying liability under the bond and in simultaneously alleging that the government, rather than Morton, had breached the construction contract. While we agree that to admit such evidence was erroneous, we conclude that the error here did not “affect the substantial rights” of the sureties. We are therefore directed by Fed.R. Civ.P. 61 to disregard the error.
1. Evidence of Bad Faith.
Central to the government’s claim of bad faith was its contention that the sureties knew at the time they denied liability under the performance bond that there was no basis for the claim that Morton had not breached the construction contract. The government sought to show such knowledge in three ways.
First, the government elicited testimony that the sureties knew, prior to Morton’s termination, that Morton’s performance had been extremely poor. For example, Mr. Calvin Persinger, one of the DOA’s contracting officers on the Plum Island project, testified that the sureties had been sent copies of the various cure notices that the government had served on Morton, He also testified that the sureties were made aware of the problems that the subcontractors were having with Morton.
Second, the government introduced evidence that the sureties’ investigation of the situation following Morton’s termination was woefully inadequate and provided no basis for the allegation that the government was in breach. Robert Kane, who at the time Morton was terminated headed Seaboard’s Smithtown office and was closely involved at the beginning with providing the bond, testified that despite his knowledge of the project, his superiors at Seaboard never consulted him on how to proceed with regard to Morton’s termination.
Finally, Seaboard was accused of instructing its consultants in the matter not to put any reports in writing, with the implicit suggestion that Seaboard officials knew that such reports would be adverse to their position and could damage the sureties’ litigation stance. See, e.g., testimony of Mark Pessolano that “Mr. Hart instructed them ... to keep notes of their investigation but not to give him a written report”, and a consultant’s memorandum that Pessolano told him “keep your own in-house notes anyway (sic) you want to but do not put it in write (sic) to Seaboard Surety.”
The stated purpose for introducing such evidence was the government’s theory that it could recover damages in excess of the bond limit of $10,685,000 if it established bad faith on the part of the sureties, and it was for this purpose that the district court admitted it. The court instructed the jury that its damages could exceed the bond limit if it found bad faith, and included in the special interrogatories question 8, asking whether the sureties had acted in bad faith.
To admit the evidence of bad faith for this purpose was error. It is hornbook law that a surety is liable up to, and only up to, the limit on the bond it issued. See, e.g., Pennsylvania Fire Insurance Co. v. American Airlines, Inc., 180 F.Supp. 239, 241 (E.D.N.Y.1960); Bill Curphy Co. v. Elliott, 207 F.2d 103, 106 (5th Cir.1953). Despite its claim of a “developing nationwide policy”, the novel exception to this rule Urged by the government is supported by little relevant precedent. The government’s citation of Transport Insurance Co. v. Michigan Mutual Liability Insurance Co., 496 F.2d 265 (6th Cir.1974), is misleading at best, since that case dealt with insurance rather than suretyship law and, more importantly, it did not hold, as the government claims, that a “surety has *964to determine if its principal has or has not breached the construction contract before it repudiates liability.” Government’s Brief at 83. Rather, it held that Michigan Mutual had not acted in bad faith, and reversed a district court finding to the contrary as clearly erroneous.
Some support for the government’s theory is provided by Continental Realty Corp. v. Andrew J Crevolin Co., 380 F.Supp. 246, 253-55 (S.D.W.Va.1974), in which Judge Merhige did find bad faith on the part of a surety, General Insurance Company. There, however, General’s refusal to perform pursuant to the bond was in disregard of an injunction. Id. at 254. We need not consider whether in such circumstances a surety’s bad faith could be considered so egregious as to render it liable for damages caused by its conduct in excess of the bond amount, since here there was no such injunction directing the sureties to perform.
The alleged bad faith of the sureties here amounted, for the most part, to no more than exercising their right to litigate the question of Morton’s breach of contract. The government alleges no separate injury attributable to the sureties’ purported bad faith (except delay in recovering its damages, which is fully compensated by the interest assessed on the judgment), but rather seeks only to recover the full quantum of damages attributable to Morton’s breach. Where all alleged damages are the result of the principal’s breach and not the surety’s conduct, we hold that evidence of the surety’s conduct is irrelevant. Hence, it was error to admit such evidence of bad faith by the sureties in this case.
2. Lack of Prejudice to the Sureties.
The standard for reversal based on the erroneous admission of evidence is set forth in Rule 61, Fed.R.Civ.P. It provides,
No error in either the admission or the exclusion of evidence and no error or defect in any ruling or order or in anything done or omitted by the court or by any of the parties is ground for granting a new trial or for setting aside a verdict or for vacating, modifying, or otherwise disturbing a judgment or order, unless refusal to take such action appears to the court inconsistent with substantial justice. The court at every stage of the proceeding must disregard any error or defect in the proceeding which does not affect the substantial rights of the parties.
It does not appear on the record before us that upholding the judgment against the sureties is “inconsistent with substantial justice”, nor did the admission of the bad faith evidence “affect the substantial rights” of the sureties.
To begin with, the evidence did not accomplish the specific erroneous result for which it was admitted, since the verdict did not exceed the penal sum of the bond. In order to show prejudice, then, the sureties must demonstrate that some element of the damages awarded by the jury — or the very finding of liability itself — was influenced by the evidence of bad faith. It is the sureties’ burden to show prejudice, Palmer v. Hoffman, 318 U.S. 109, 116, 63 S.Ct. 477, 481, 87 L.Ed. 645 (1943) (“He who seeks to have a judgment set aside because of an erroneous ruling carries the burden of showing that prejudice resulted.”); In re Air Crash Disaster at John F. Kennedy International Airport on June 24, 1975, 635 F.2d 67, 70-71 (2d Cir.1980), but the sureties have failed to sustain their burden. The strongest point they raise is that the jury obviously was persuaded by the evidence of bad faith since it answered “Yes” to question 8 of the special interrogatories that comprised its verdict, “Did the Sureties act in bad faith?” But standing alone, the fact that the jurors were persuaded by the evidence does not mean that the evidence influenced their deliberations on the other issues in the case.
The key questions were whether the government or Morton had breached the construction contract, and, if Morton had done so, what the government was entitled to recover in damages. By far the vast majority of the 1,700 pages of trial transcript is devoted to the issue of which party had breached. In answer to question 2 of the special interrogatories, “Did Mor*965ton Company breach its agreement with the Government?”, the jury answered “Yes.” There was overwhelming evidence that Morton did in fact breach its construction contract with the government, and that the government thereby incurred millions of dollars in damages, spent in assessing the task of completion left to the government by Morton, repairing the poor work done by Morton, and, only then, actually completing the three buildings that available funding permitted. A fair reading of the record supports the jury’s verdict as to Morton’s breach; in fact, the sureties do not seem to be contending that the improperly admitted evidence influenced the outcome on that issue. Rather, they focus on damages.
While it is impossible to discern from the record precisely what formed the basis for the amount of damages awarded by the jury, it is clear that the jury did not simply adopt uncritically the claims of the United States. The government was asking for $11,420,000 in damages, but the jury awarded $10,631,326, or some $788,674 less. This supports the notion that the jury did in fact examine the damages claimed by the government, and gave judgment based on that examination rather than on the basis of its view of the sureties’ conduct. Presumably, if the jurors wished to punish the sureties for their bad faith they would have awarded the government at least as much as, and perhaps more than, the amount it was seeking. Of course, it is conceivable that the jury would have disallowed even more of the damages sought by the government if not for the bad faith found, but such a supposition is supported only by speculation.
The sureties seek to show that the damage award was affected by the evidence of bad faith by arguing that the jury failed to give them any credit for the value of work done by Morton immediately prior to termination and as reflected in “change orders” submitted by Morton that were pending when Morton was terminated. The sureties contend that the government “conceded” that the value of that work was $300,000, and that the failure of the jury to credit the sureties for at least that amount must be the result of the finding of bad faith.
This argument is, at best, strained. The sureties point to no logical connection between the change orders and the bad faith, and no such connection appears to exist. Moreover, the government’s “concession” as to the change orders was accompanied by a statement that, even assuming the $300,000 value of that work, Morton had still been overpaid for the work it did on the entire project. In any event, the argument that the jury’s refusal to credit the sureties with $300,000 in change orders is evidence of prejudice amounts to speculation, not proof or even plausible argumentation.
In sum, there is no claim by the sureties that the evidence of bad faith on their part in any way influenced the jury’s conclusion that Morton, and not the government, had breached the contract. Since the government offered no proof — indeed, raised no argument — of any damages except those caused by Morton’s breach, there is no reason to believe that the amount of the damages awarded was due to anything other than that proof and that breach. The sureties have established no element of the damages that resulted from or was influenced by the erroneously admitted evidence. In short, “we cannot say that the ruling was prejudicial even if ... it was erroneous.” Palmer v. Hoffman, 318 U.S. at 116, 63 S.Ct. at 482. See also Bickerstaff v. S. Central Bell Tel. Co., 676 F.2d 163, 169 (5th Cir.1982).
C. Prejudgment Interest.
The sureties finally contend that the district court erred in the amount of prejudgment interest it added to the damages found by the jury. We agree with the sureties that the district court’s calculation of interest was incorrect, but not for the reasons advanced by the sureties. No statute or regulation governs the award of interest on a judgment based on a Miller Act bond, and therefore the entry of prejudgment interest is committed to the sound discretion of the district court. The *966court’s decision here was, however, an abuse of discretion.
The sureties’ argument that interest should have run from January 6, 1986, the date the current contracting officer issued a formal demand for payment once the completion work was finished, requires little comment. They contend that interest determinations are governed by the Federal Acquisition Regulations (“FARs”), 48 CFR § 1 et seq., which require a demand for payment that shall include notice that interest will be assessed on any part of the demand remaining unpaid after thirty days. 48 CFR § 32.610(b)(2). The FARs, however, like the CD A, govern a debt incurred pursuant to a government contract, which a Miller Act performance bond is not, and specifically to a-contract for the acquisition of supplies and services, which as we noted in § A(l) above, a bond is certainly not. We see no more reason to expand the scope of the FARs to reach Miller Act bonds than we did similarly to expand the CDA.
The calculation of prejudgment interest was therefore committed to the discretion of the district court. Rodgers v. United States, 332 U.S. 371, 373, 68 S.Ct. 5, 6, 92 L.Ed. 3 (1947); EEOC v. County of Erie, 751 F.2d 79, 81 (2d Cir.1984). However, the purpose of prejudgment interest is compensatory, not penal. See Rodgers, supra (a party “who has suffered actual money damages by another’s breach ... should be fairly compensated” by entry of appropriate interest). Here, the quantum of interest assessed by the district court goes well beyond fairly compensating the government for the loss it sustained by virtue of the fact that the sureties denied liability on the bond.
The penal quality of the district court’s award becomes clear upon consideration of what would have occurred had the sureties conceded, rather than denied, liability in 1980 and agreed to pay the completion costs. They would have had to reimburse the government as those costs were incurred; simultaneously, however, they would have had the full use of — and benefited fully from the interest obtainable on — the money until it was actually paid to the government. Thus, assessing interest from the day the sureties denied liability— even if that action is viewed as something akin to an “anticipatory breach” rather than simply an assertion of the sureties’ right to litigate their liability — penalizes them instead of merely compensating the government.
From the government’s perspective, it was harmed only insofar as it was forced to pay for completion that should have been financed by proceeds from the bond. Interest, therefore, cannot be assessed prior to the date the government began paying to complete the Plum Island project. Moreover, after it terminated Morton, the government had $2,579,704 in appropriated funds that remained undisbursed from the original contract, money that it used to pay the initial costs of completion. This the government would have been obligated to do even had the sureties agreed to pay on the bond. Thus, the government was not legally or actually damaged by the sureties’ failure to accept liability until its expenditures on completion exceeded $2,579,704. The sureties contend that that occurred in January 1983, but on remand the district court is free to hear argument and accept evidence on that question.
We recognize that a dollar-for-dollar, check-by-check calibration of the damages awarded by the jury with the government’s completion expenditures is impossible. Since the verdict disallowed some of the claimed damages, and since it is impossible to discern which of the costs formed the basis for the award, there is no way to tie prejudgment interest to the exact date the government spent each dollar toward completion.
Fortunately, the district court need not, on remand, attempt the impossible: an exact calculation of payments on which interest would be due. The court has considerable discretion in assessing interest, and by our discussion of the framework that should guide its analysis we do not intend to suggest that the interest on each dollar of the verdict must be assessed only from *967the day it can be shown the government spent that dollar towards completion. So long as the district court’s award of interest generally reflects a level that roughly and fairly compensates the government for those costs, it will fall within the range of the court’s discretion. We therefore remand for recalculation of prejudgment interest beginning no earlier than the date the government’s post-default expenditures toward completion of the Plum Island project exceeded $2,579,704.
Affirmed in part, reversed in part, and remanded.