The United States appeals a pre-trial order barring three of defendant's alleged accomplices from testifying against him. The district court's decision to exclude their testimony was based on its conclusion that the plea agreements entered into by the accomplices were so likely to induce perjurious testimony that to allow them to testify would be to violate defendant's due process rights. United States v. Dailey, 589 F. Supp. 561 (D.Mass.1984). Finding ourselves in disagreement with the district court, we reverse.

On November 25, 1983, a federal grand jury for the District of Massachusetts returned two major drug smuggling indictments. In the first indictment defendant Kevin R. Dailey is named as one of two lead defendants. In Count One, he is charged as an organizer, supervisor, and manager of a continuing criminal enterprise to violate the controlled substances laws, 21 U.S.C. Sec. 848, and in particular with helping to organize a number of drug smuggling ventures involving an aggregate of approximately 185,000 pounds of marihuana. The remaining counts of the indictment charge Dailey, as well as several others,1 with specific drug violations based upon certain of the transactions cited as predicates in Count One.

The lead defendant named in the second indictment is Salvatore Michael Caruana. In Count One he is charged under section 848 as an organizer, supervisor, and manager of a second continuing criminal enterprise to import and distribute large quantities of marihuana. The remaining counts of the indictment charge him and several others with a number of specific drug offenses.

The Dailey and Caruana indictments are related and have common counts. Dailey's enterprise allegedly worked with Caruana's on a number of drug ventures, including: (1) the importation of approximately 90,000 pounds of marihuana into Greenleaf Cove, Maine, and other ports in or about May 1979; and (2) the importation of approximately 45,000 pounds of marihuana into Stonington, Maine, and other ports in or about September 1979. Both the Dailey and Caruana indictments include counts covering these two ventures. Also involved in these ventures was a third enterprise, one run by Robert L. Frappier and Timothy I. Minnig. They and one of their subordinates, Tommy Tindall, are the three accomplices whose plea agreements are the subject of this appeal.

Ten of the defendants charged in the Caruana indictment were tried during May and June of 1984 and eight of them were found guilty.2 An important subject of that trial was the September 1979 venture. Among the several witnesses who testified for the government were Frappier and Minnig, who had agreed to cooperate with the government four months after Dailey and Caruana were indicted, and Tommy Tindall, who had agreed to cooperate a year-and-a-half earlier. After the trial was underway and Frappier, Minnig, and Tindall had testified, defense counsel moved to strike their testimony, making the same basic argument that Dailey advances here. Counsel argued that certain promised benefits to the accomplices were, by the terms of their plea agreements, made contingent upon the successful prosecution of the accomplices' confederates. The contingent nature of these benefits, it was contended, had provided such an inducement for the accomplices to testify falsely that their testimony should be stricken from the record.

This argument was made after the accomplices had testified because it was not until then that any federal case authority in support of the argument had come to defense counsel's attention. When, however, the Court of Appeals for the Eighth Circuit published its initial panel decision in United States v. Waterman, 732 F.2d 1527, 1531 (8th Cir.1984), later to be vacated en banc, id. at 1533, counsel had one case standing for the proposition that a plea agreement made contingent upon the return of further indictments constitutes a violation of the due process rights of any defendant against whom the confederate who entered the agreement might testify.

In Caruana, although the district court carefully considered Waterman, it denied defendants' motions to strike. While noting that the plea agreements were somewhat unusual, the court held that the standard procedural safeguards--having the agreements read to the jury, allowing defense counsel to cross-examine the witnesses with respect to the agreements, and instructing the jury to scrutinize the accomplices' testimony with care--were sufficient to protect defendants' due process rights. In ruling on the motion, the court found compelling the existence of "corroborating evidence in the form of repair records, telephone records, [the] police testimony and [another] witness's testimony." District Judge Zobel commented:

"I credit, in general, the testimony of Messrs. Tindall, Frappier [and] Minnig.... Such discrepancies as there were in details are minor and they do not, in my judgment, substantially affect their stories which are remarkably consistent, both internally and with respect to each other. I do not find the testimony of ... these people ... blurred by the plea agreements which they had entered."

Three days later, however, just before Dailey's trial was to commence, District Judge Tauro considered the same plea agreements and came to the opposite conclusion. United States v. Dailey, 589 F. Supp. at 564-65. Relying exclusively upon the original panel decision in Waterman, he found that use of the three accomplices' testimony would violate Dailey's due process rights. Accordingly, he barred the three from giving any testimony, including testimony concerning the same September 1979 venture about which they had already testified in the Caruana trial. Id. It is this pre-trial exclusionary ruling that the government now appeals.

Frappier and Minnig signed virtually identical written plea agreements. They cover many drug smuggling ventures and charges brought in Maine, Massachusetts, and Oregon. Paragraph 5 of each agreement, the primary provision at issue on appeal, reads as follows:

"The defendant agrees to fully cooperate, as defined in Paragraph 2. If, at the time of sentencing on the Maine indictment, the defendant has fully cooperated with the United States, as defined in Paragraph 2, the Government will recommend a specific term of imprisonment which does not exceed twenty (20) years and, depending principally upon the value to the Government of the defendant's cooperation, the Government, in its sole discretion, may recommend a sentence of ten (10) years; the defendant may argue for a sentence less than the Government's recommendation; in any event, the Court shall impose a sentence no greater than that recommended by the Government. If, at the time of sentencing on the Maine indictment, the Government presents evidence and the Court finds by a preponderance of the evidence that the defendant has not fully cooperated, as defined in Paragraph 2, then the court shall sentence the defendant to a term of imprisonment of thirty-five (35) years. The Court shall not sentence the defendant to pay a fine on either the Maine or Oregon indictment."

"Full cooperation" is defined in Paragraph 2 of each agreement as: (1) the disclosure of all criminal activities known to the defendant, whether undertaken by the defendant or others; (2) the provision of all records, receipts, etc., which tend to corroborate the occurrence of such criminal activities; and (3) the giving of "complete and honest testimony at any and all proceedings" concerning such activities.

The district court read Paragraph 5 to mean that failure to fully cooperate would result in a sentence of 35 years, complete and honest cooperation would result in a sentence of 20 years, and cooperation that is of "value" to the government, or, as the district court rephrased the matter, which results in "the success of the underlying prosecution", would result in a sentence of 10 years. 589 F. Supp. at 564. So viewing the contingent nature of the ultimate sentence, the district court found that the plea agreements on their face impose a subjective pressure on the accomplices to testify as the government wants rather than truthfully and that the creation of such an inducement to lie necessitates excluding all their testimony. Id.

On appeal, the government notes that the plea agreements do not make the ultimate sentences contingent upon the prosecution or conviction of anyone, and that neither Frappier nor Minnig, according to their testimony in the Caruana trial, understood the agreements to intend such a contingency. In addition, the government disagrees with the district court's isolation of the "value to the Government" provision from the rest of the agreement. The government contends that what was intended and understood by the parties was that this provision effectively promised that "complete and honest" cooperation would result in a sentence which could range from ten to twenty years, depending upon the extent and quality of the information that Frappier and Minnig provided.

Among the circumstances cited by the government as requiring it to set a range of years rather than choose a specific number are the following: the fact that, although Frappier and Minnig were very high level drug smugglers and probably knew a great deal, it was impossible to determine the extent of their knowledge before they signed plea agreements and felt free to talk; the fact that what they knew and could honestly report had to be factored into the ultimately recommended sentence in order for both the public and the pleading parties to obtain a just and proper quid pro quo in the exchange of a specific reduction in sentence for information of a particular value in terms of achieving valid law enforcement ends; and the fact that the government anticipated a number of prosecutions in a number of jurisdictions extending over a number of years, which created a need to leave the ultimate sentence somewhat open and dependant upon the extent to which Frappier and Minnig continued to cooperate over time.3 The government contends that if the agreements are viewed in this light, they cannot be seen as creating an impermissible inducement to lie.

For a number of years Tindall worked for Frappier and Minnig as a pilot, offloader, and general assistant. Arrests for some of his drug-related activities led to three convictions in 1981 and 1982. Subsequently faced with sentences totalling five and one-half years of imprisonment and the possibility of further prosecution, Tindall agreed to cooperate with the government in August 1982.

In exchange for divulging his knowledge of illegal drug activities, Tindall was promised a four-month stay of sentencing, the possibility of a further stay, and the possibility of government support for any motion for reduction of sentence that Tindall might later file. The latter two benefits were made contingent upon the value or "benefit" of his information to the government. The district court stated that the contingent provisions in Tindall's agreement was impermissible because "[t]he pressure to please under such an arrangement is potentially coercive." 589 F. Supp. at 564. On appeal, the government attacks the district court's exclusion of Tindall's testimony on the same grounds that it attacks the Frappier and Minnig rulings.

After carefully considering the plea agreements themselves, the testimony given at the Caruana trial, and the law governing the admissibility of accomplice testimony, we conclude that the risk of perjury created by the agreements is not so great that Dailey's due process rights will be violated by the admission of the accomplices' testimony at trial. While we share the concern and uneasiness of the district court over the coercive potential of these plea agreements, we believe that the traditional safeguards employed in the Caruana trial and discussed below should adequately protect Dailey's rights.

We begin by noting that the initial decision in Waterman, 732 F.2d at 1531 was unprecedented in modern times. Long ago the courts rejected the notion that the testimony of co-defendants and other interested witnesses was so likely to be unreliable that it should be excluded. See, e.g., Benson v. United States, 146 U.S. 325, 13 S. Ct. 60, 36 L. Ed. 991 (1892); Rosen v. United States, 245 U.S. 467, 38 S. Ct. 148, 62 L. Ed. 406 (1918). Recognizing that such individuals were frequently the most knowledgeable witnesses available, the courts have chosen to allow them to testify and to rely upon cross-examination to ferret out any false testimony they might give. As the Supreme Court put the matter in Hoffa v. United States, 385 U.S. 293, 311, 87 S. Ct. 408, 418, 17 L. Ed. 2d 374 (1966), "[t]he established safeguards of the Anglo-American legal system leave the veracity of a witness to be tested by cross-examination, and the credibility of his testimony to be determined by a properly instructed jury." In the particular case of an accomplice who has struck a plea agreement, the "established safeguards" are that the jury be informed of the exact nature of the agreement, that defense counsel be permitted to cross-examine the accomplice about the agreement, and that the jury be specifically instructed to weigh the accomplice's testimony with care. See, e.g., United States v. Insana, 423 F.2d 1165, 1169 (2d Cir.1970).

Given this state of the law, it is not so surprising that the Eighth Circuit, sitting en banc, vacated the initial Waterman holding that a trial court violates a defendant's due process rights by admitting the testimony of an accomplice who has entered into a contingent plea agreement with the government.4 Id. at 1533. As a consequence of the vacatur, the district court's ruling in this case, which depended exclusively upon the initial Waterman decision, is left without any case support.

Moreover, even if the original Waterman decision still stood, we do not think its holding would apply in this case, which we find to be factually distinguishable. In Waterman, 732 F.2d at 1530, the plea agreement included a government promise to recommend a two-year reduction in the accomplice's sentence if his "truthful testimony led to further indictments...." Here, by contrast, the ultimate sentence to be recommended is dependent upon something rather indefinite, i.e., the "value" or "benefit" of the accomplice's cooperation to the government, rather than upon the very specific occurrence of further indictments. Although the district court characterized the contingency involved here as the achievement of successful prosecutions, we note that the agreements do not say this on their face.5 Nor did any of the accomplices testify under cross-examination in the Caruana trial that they thought the agreements created such a contingency. Accordingly, we find that the district court lacked sufficient evidence to characterize the critical contingency as the occurrence of successful prosecutions.

In addition, we note that the government has established fairly compelling reasons for having made these plea agreements somewhat open-ended. Frappier and Tindall, in particular, were strongly suspected of being major organizers and financiers of multi-ton drug importation schemes. The chances were excellent that they honestly knew enough to support a number of prosecutions in a number of jurisdictions against many individuals involved in many schemes. As a result the government had a legitimate interest in striking a plea bargain certain enough to induce these individuals to plead and start cooperating but also contingent enough upon what they knew and how fully they cooperated over a period of time to persuade them to continue their cooperation until the fund of knowledge they held had been completely tapped and utilized in the investigation and prosecution of their alleged confederates.

Finally, there is a need to correct the district court's finding that the "value to the government" provision of the Frappier and Minnig agreements is unrelated to the other provisions. 589 F. Supp. at 563-64. A close reading of the agreements reveals that it is not necessary to interpret them as promising a twenty-year term for honest testimony and a ten-year term for valuable testimony, whether honest or not. Paragraph Two of the Frappier and Minnig agreements explicitly defines "full cooperation" as requiring, among other things, the giving of "complete and honest testimony". Paragraph Five states that failure to provide full cooperation will result in a sentence of 35 years. Lying, therefore, appears to carry with it a risk of being caught in the lie and consequently receiving 35 years for failing to give honest testimony, rather than ten years for providing valuable information.6 This risk is greatest, of course, with respect to testimony that can be contradicted by the testimony of others or by documentary evidence, a great deal of which was presented in the Caruana case and will be presented in this case.

Although we are hesitant to make too much of this point, we think it demonstrates the fallacy in the district court's finding that the key contingency provisions in these agreements necessarily refer to something other than full and honest cooperation. While we do not doubt that the agreements could have been drafted to make clearer the fact that "negative" value would attach to any perjurious testimony and while we think the government should do just that in the future, we reject the implication that these agreements on their face betray an impermissible overreaching by the government.

To summarize, while these agreements are somewhat comparable to that found in Waterman, they differ significantly in that no promised benefit is made contingent upon a resulting indictment or conviction. Thus, while the district court was correct in concluding that these agreements provide some inducement to lie or to embellish the facts, the question is whether that inducement is significantly greater in degree or more sinister in quality than the inducements created by agreements that have passed judicial muster in the past.

The general rule allowing an accomplice who has pled guilty to testify certainly recognizes that a risk of perjury is created by the plea agreement itself. In United States v. Meinster, 619 F.2d 1041, 1045 (4th Cir.1980), for example, the court declared it "obvious that promises of immunity or leniency premised on cooperation in a particular case may provide a strong inducement to falsify in that case." Nonetheless, the jury is entrusted with these matters even in cases where there is a plea agreement and the accomplice's background is most questionable. Thus, for example, in United States v. Miceli, 446 F.2d 256, 258-59 (1st Cir.1971), we held that it was proper to admit the uncorroborated testimony of an accomplice who had a long list of convictions, was a thief by profession, had committed perjury in another case, and was the beneficiary of a generous plea agreement with the government. As another court stated in general terms, "a conviction may be based on the uncorroborated testimony of an accomplice ... even though the accomplice is in a position to gain favors from the government by his testimony ... and even though there are inconsistencies in his story ... so long as it is not 'incredible or unsubstantial on its face.' " Lyda v. United States, 321 F.2d 788, 794-95 (9th Cir.1963).

Courts have followed this rule even when confronted with an agreement that leaves an accomplice's sentence open until after he has testified. In United States v. Insana, 423 F.2d at 1168-69, for example, the court rejected a due process challenge to the admission of testimony from two accomplices who had yet to be sentenced. The court stated that "the mere fact that a witness hopes to receive a reduced sentence or some other form of leniency does not disqualify him as a witness but affects only the weight of his testimony." Id. at 1168; see also United States v. Borman, 437 F.2d 44, 45-46 (2d Cir.1971) (where jury was properly instructed and credibility issue fully argued, trial court committed no error by allowing testimony of unsentenced accomplice who had pled guilty). So firmly established is the policy of allowing unsentenced accomplices to testify that we find one court, in rejecting the exclusion of one such accomplice's testimony, declaring, "We are not disposed to fashion a rule heretofore unknown in the jurisprudence of criminal prosecutions. Neither are we impressed that the testimony of an unsentenced accomplice deprives one who stands trial of due process or fair treatment." United States v. Vida, 370 F.2d 759, 767 (6th Cir.1966).

Even more directly on point is United States v. Kimble, 719 F.2d 1253 (5th Cir.1983). There an accomplice pled guilty and agreed to testify with the understanding that he would be sentenced to 20 years imprisonment if his cooperation was considered adequate by the government. Approving the admission of his testimony after the fact, the appellate court explained:

"[The accomplice] was subjected to rigorous cross-examination, including extensive questioning about the precise nature of his plea arrangement, his personal background and his crime-studded record. [He] acknowledged that he had repeatedly lied, including lies under oath, to save his own skin. He admitted to intimate participation in the criminal activity set forth in the indictment and candidly stated that he was testifying only because of the lenient sentence he was to receive in return for his cooperation. In particular he spoke of the contingent nature of his sentence, which had not yet been imposed, and that he fully understood his cooperation would have to be deemed acceptable before the agreement would be operative." Id. at 1255 (emphasis added).

The accomplice there was every bit as likely to give unreliable testimony as are the accomplices here. The nature of the contingency in his agreement was very similar to that found in the agreements of Frappier, Minnig, and Tindall: there the contingency was vaguely defined as the "acceptability" of the accomplice's cooperation; here it is the "value" or "benefit" of such cooperation. Yet, the Kimble court refused to hold that the accomplice's testimony was per se incredible or that it should have been excluded. Instead, the court ruled that the plea agreement, the accomplice's admitted prior lying, and his prior criminal record all went to the weight of his testimony, not to its admissibility. Id. at 1257; see also United States v. Evans, 697 F.2d 240, 245 (8th Cir.1983).

Similarly, courts have allowed the testimony of informants who agree to work undercover and testify in exchange for either leniency, money or both. For example, in United States v. Tapia, 738 F.2d 18, 20 (1st Cir.1984), we upheld the admission of testimony from an informer who had agreed to undertake an unmonitored "controlled buy" of drugs and to testify about that buy in exchange for a government promise concerning leniency. The risk of developing unreliable evidence there was easily as great as the risk of obtaining unreliable testimony here; yet, we ruled that the risk was not great enough to require anything more than the standard procedural safeguards. Id. Likewise, in United States v. Valle-Ferrer, 739 F.2d 545, 549 (11th Cir.1984), the court held that an informant's anticipated receipt of $1,000 if his testimony resulted in a conviction did not render him incompetent to testify. In Heard v. United States, 414 F.2d 884, 886 (5th Cir.1969), the court approved the admission of testimony from an informant who was a convicted felon and who had been paid for results rather than for information. Finally, in United States v. Crim, 340 F.2d 989, 900 (4th Cir.1965), the testimony of undercover agents was found admissible even though the amount of their compensation was to be determined after trial "on the basis of an appraisal of the extent and quality of [their] work...."

Although Dailey cites Williamson v. United States, 311 F.2d 441 (5th Cir.1962), as an analogous contingent fee case in which the conviction was overturned, Williamson has no applicability here. It involved an agreement under which the agent was to be paid $200 for successfully helping the government catch a particular person in the act of committing a crime which was to be set up, at least in part, by the agent. The case presented a serious question of entrapment, which is not present in this case. The court was disturbed by the fact that the agreement called for the conviction of a particular person even though the government appeared to lack any grounds for believing that the person was engaging in criminal activity. Id. at 444. In subsequent cases, the Fifth Circuit has carefully limited Williamson to its facts. See, e.g., United States v. Gray, 626 F.2d 494, 499 (5th Cir.1980) (testimony of two informants allowed even though they received $37,000 and $25,000, respectively, since payment was not made "contingent upon conviction of a particular person"); United States v. Edwards, 549 F.2d 362, 365 (5th Cir.1977) (allowing testimony of informant whose receipt of a reward was made contingent upon the "final results" of his undercover work); Hill v. United States, 328 F.2d 988, 989 (5th Cir.1964) (Williamson applies only if government lacks prior knowledge of suspect's unlawful activity). In United States v. Jett, 491 F.2d 1078, 1081 (1st Cir.1974), we too distinguished Williamson, stating that payment of an informant on the basis of the value of the services rendered is a permissible contingent fee arrangement.

The Sixth Circuit has gone even further. In United States v. Jones, 575 F.2d 81, 85-86 (6th Cir.1978), it explained its rationale for refusing to follow Williamson by quoting extensively from an earlier decision:

" '[There is no overriding policy to exclude the testimony of an informant if he is paid under] a contingent fee agreement for the conviction of specified persons for crimes not yet committed. Although it is true that the informant working under this type of arrangement may be prone to lie and manufacture crimes, he is no more likely to commit these wrongs than witnesses acting for other, more common reasons.... Rather than adopting an exclusionary rule for a particular factual situation, ... we prefer the rule that would leave the entire matter to the jury to consider in weighing the credibility of the witness-informant. (Citation omitted). In our view this approach provides adequate safeguards for the criminal defendant against possible abuses since the witness must undergo the rigors of cross-examination.' " Id. at 86 (quoting United States v. Grimes, 438 F.2d 391, 395-96 (6th Cir.), cert. denied, 402 U.S. 989, 91 S. Ct. 1684, 29 L. Ed. 2d 155 (1971)).

Whether or not we would go as far as the Sixth Circuit with respect to the testimony of informants, we find ourselves in agreement with the general policy and reasoning of Jones. The agreements in the instant case and the risk of perjury that they entail are not significantly different from or greater than those described in the cases above. There is therefore no reason to treat these accomplices any differently than were the accomplices and informants in those cases.7

Accordingly, we hold that Frappier, Minnig, and Tindall should be allowed to testify against Dailey. The standard procedural safeguards that were employed in the Caruana proceedings should be employed here as well. The agreements should be read to the jury and made available during its deliberations; defense counsel may cross-examine the accomplices at length about the agreements; and the jury should be given the standard cautionary instruction concerning the testimony of accomplices and a special cautionary instruction concerning the nature of each accomplice's contingent agreement and the risk that it creates, particularly in instances where the accomplice's testimony cannot be corroborated.8

Finally, with respect to any future plea agreements that the government may strike, we suggest that they should contain a clear statement that the giving of false information or testimony will be considered a failure to fully cooperate and will vitiate any contingent provision in the agreement. In addition, we recommend that such agreements specify as precisely as possible those factors which the government is to take into accountin determining the degree to which the contingent condition has been satisfied. It is the government's duty to evaluate carefully whether and why a contingency agreement is necessary, as well as to insure that all parties have a clear understanding of what each contingency provision means and what it does not mean. Also, we note that at present we can think of no instance in which the government would be justified in making a promised benefit contingent upon the return of an indictment or a guilty verdict.9 Finally, we note that contingency agreements should be reserved for exceptional cases, such as this one, where the value and extent of the accomplice's knowledge is uncertain but very likely to be great.

The district court's decision to exclude the testimony of the three government witnesses is vacated and the case is remanded for further proceedings in accordance with this opinion.