OPINION

WILKINSON, Circuit Judge:

Ervin Blevins, Dewey Johnson, and Joel Mitchell appeal their convictions for participation in a drug distribution conspiracy and other drug-related offenses. These convictions came at the end of a complex seven-day trial. As the bases for reversal of the convictions, appellants now allege fourteen separate grounds of error. While we reject thirteen of these grounds as being without merit, we conclude that the introduction into evidence of testimony that six non-testifying co-defendants had pleaded guilty constituted error.

This error must be subjected, however, to harmless error analysis, which must be undertaken with care in a complex drug conspiracy case. Typically such multiple defendant cases present upon appeal difficult questions involving, inter alia, the admission of coconspirator statements, the scope and duration of the alleged conspiratorial conduct, and the introduction of confessions of non-testifying co-defendants that arguably implicate other defendants in the case. The dangers of unreliable or unfairly prejudicial evidence in such cases are significant, and courts must be alert to them. At the same time, these cases may present a technical minefield for prosecutors and judges which makes the perfect trial more the exception than the rule. The defendants in this case were not given a perfect trial, but they were given the fair trial to which they were entitled under the Constitution. We are left, after a careful review of the record, with the firm conviction that the error in this case was harmless beyond a reasonable doubt. We, therefore, affirm the convictions.

The criminal prosecutions here stemmed from an undercover drug investigation conducted by the Drug Enforcement Administration in the Saltville area of southwest Virginia from May 1989 to March 1990. Acting on reports of methamphetamine and cocaine trafficking in the area, the DEA assigned Special Agent David High as its chief undercover operative in the Saltville probe. He was assisted throughout the investigation by a paid DEA informant known as Bear. High began the investigation by making small purchases of drugs from relatively minor dealers in the hope of eventually discovering the major distributors in the area.

Over the first few months of the operation, High purchased drugs from a number of individuals, and through these contacts, he became aware that Joel Mitchell occupied a prominent role in the drug trade in the area. When High later learned that Mitchell had been having difficulty in obtaining a cheap supply of cocaine, High shifted to a reverse undercover operation in which he and Bear portrayed themselves *1255 as drug suppliers rather than simply small scale purchasers.

The initial direct contact between High and Mitchell took place on February 16, 1990. At this meeting, High made clear that he could supply Mitchell with cocaine and marijuana. After a series of discussions and negotiations over the next few weeks, Mitchell agreed on March 2, 1990, to purchase three kilograms of cocaine and at least 35 pounds of marijuana from High and Bear. At Mitchell’s request, High conducted a practice air drop on March 13, 1990, two days before the scheduled delivery of the drugs. On March 15, 1990, a team of law enforcement officers arrested Mitchell when he arrived at the site where the drugs were to be dropped.

Ervin Blevins and Dewey Johnson were also arrested on March 15. Blevins had been present at Mitchell’s house on March 2 when the purchase negotiations were finalized, and he accompanied Mitchell, High, and Bear on a tour of Mitchell’s property later that day to scout potential sites for the air drop of the drugs. Blevins was also with Mitchell on March 15 at the site of the planned drop, where Blevins was arrested. As to Dewey Johnson’s involvement, Mitchell had made statements to High on March 13 and 15 indicating that he had a partner for the drug purchase. Dewey Johnson matched the description of the unnamed partner, and Johnson was arrested when he was found where Mitchell said his partner would be waiting on the day of the drug delivery.

A federal grand jury returned a twenty-seven count indictment against Mitchell, Blevins, Johnson, and eight other individuals on March 23, 1990. The indictment charged all eleven defendants with participation in a drug trafficking conspiracy and with various other individual crimes related to the drug activities. Prior to trial, eight coindictees entered guilty pleas, leaving only appellants to be tried.. The trial of appellants lasted seven days. During the course of the proceedings, the district court entered a judgment of acquittal at the close of the government’s case in favor of Joel Mitchell on the charges that he had engaged in a continuing criminal enterprise and money laundering and in favor of Dewey Johnson on a money laundering charge. The court submitted the remaining counts to the jury, which returned verdicts of guilty on all of them.

II.

Three assignments of error involve challenges to the application of Federal Rule of Evidence 801(d)(2)(E), which defines as non-hearsay statements made by a co-conspirator during the course of and in furtherance of a conspiracy. We find these three contentions to be without merit.

Appellant Johnson claims that the district court erred in admitting co-conspirator Joel Mitchell’s statements against him pursuant to Rule 801(d)(2)(E). Those statements, as related by Agent High, detailed Johnson’s role in funding the drug purchases and also described his physical characteristics and the circumstances of a prior arrest involving significant quantities of marijuana. To admit evidence as a co-conspirator’s statement, a court must conclude (1) that there was a conspiracy involving the declarant and the party against whom admission of the evidence is sought and (2) that the statements at issue were made during the course of and in furtherance of that conspiracy. Bourjaily v. United States, 483 U.S. 171, 175, 107 S.Ct. 2775, 2778-79, 97 L.Ed.2d 144 (1987). Contrary to Johnson’s assertion, a court in deciding these factual questions is not precluded from examining the out-of-court statements sought to be admitted; the court may consider those statements in conjunction with other evidence. Id. at 181, 107 S.Ct. at 2781-82. These preliminary questions are subject to a preponderance of the evidence standard of proof. Id. at 175, 107 S.Ct. at 2778-79. Any findings of fact with respect to these questions are subject to a clearly erroneous standard of review, id. at 181, 107 S.Ct. at 2781-82, and a district court’s decision to admit evidence under Rule 801(d)(2)(E) may only be overturned on appeal if it constituted an abuse of discretion.

*1256 See United States v. Curro, 847 F.2d 325, 328 (6th Cir.1988).

We find no such abuse of discretion here. Johnson’s claim that the government offered no proof of his involvement in the conspiracy except for Mitchell’s statements is simply incorrect. The government offered numerous pieces of evidence that linked Johnson to the description of the unnamed partner contained in Mitchell’s statements. To take but one example, Johnson was found on March 15 at Mitchell’s vacant store at precisely the time the drug delivery was to occur. Johnson’s physical description matched that given by Mitchell. Johnson was found with $6400 on his person, and an additional $35,500 was found hidden in the vacant store. These two amounts added up to just over half of the money for the drug purchase— the amount that Mitchell indicated that “his partner” would be providing. When considered in conjunction with Mitchell’s out-of-court statements regarding his partner, the independent pieces of evidence provided ample basis for the district court to conclude that the existence of a conspiracy between Johnson and Mitchell had been shown by a preponderance of the evidence. The fact that the evidence served to corroborate Mitchell’s statements does nothing to undermine its independent status under Bourjaily. Given that Mitchell’s statements were plainly made during the course of and in furtherance of the conspiracy, we find no error in the district court’s admission of them.

Johnson also contends that a trial court may not admit evidence under Rule 801(d)(2)(E) without first making a finding on the record that all the requirements for admissibility under that rule have been satisfied. He claims that no such finding was made in this case. Johnson filed a motion in limine seeking to exclude Mitchell’s statements as evidence not falling within Rule 801(d)(2)(E). At a pretrial conference on July 2, 1990, and at the trial on July 3, the district court considered Johnson’s motion to exclude the statements. Though the judge declined to make a final ruling, he pointed out that the statements at issue in this case seemed very similar to the ones whose admission had been approved by the Supreme Court in Bourjaily. Johnson’s counsel did not object at the time Agent High testified to Mitchell’s statements. In connection with a motion for acquittal at the close of the government’s case, however, Johnson’s counsel renewed his challenge to the statements being admitted into evidence. The judge indicated at this point that he was satisfied that the requirements of Bourjaily had been met.

We find nothing improper about this procedure. This circuit has rejected the formalistic requirement that there must be a hearing to determine the existence of a conspiracy before statements can be admitted under Rule 801(d)(2)(E). United States v. Hines, 717 F.2d 1481, 1488 (4th Cir.1983). Instead, we allow a trial court to conditionally admit co-conspirators’ statements subject to the subsequent satisfaction of the requirements for their admission. Id. Moreover, we do not believe that a trial judge is required to set out on the record his reasons for making this evidentiary ruling, see Curro, 847 F.2d at 328, although it may be advisable for the trial court to do so in order for an appellate court to more easily assess the basis of the ruling. An appellate court, however, may affirm a judgment where the record reveals that the co-conspirator’s statements were plainly admissible, whether or not a detailed rationale for admitting the statements has been stated by the trial court. As noted above, the independent evidence of Johnson’s involvement in the conspiracy was substantial, and the absence of formal findings on the elements of admissibility of the Mitchell statements does not render their introduction suspect. 1

*1257 C.

All three appellants challenge the admission of Agent High’s testimony recounting statements made by Wesley Mitchell and Jerry Mitchell, the two co-indictees who testified at trial. Citing United States v. Bolick, 917 F.2d 135 (4th Cir.1990), appellants allege that this testimony by High should have been disallowed as an improper attempt by the government to introduce prior consistent statements of these witnesses before they had been impeached.

In Bolick, the statements that were improperly introduced did not qualify for admission under Rule 801(d)(2)(E) because they were made after the alleged conspiracy had ended. Id. at 137. In a case where statements are admissible under Rule 801(d)(2)(E), they may be introduced at any point, regardless of whether the credibility of the co-conspirator declarant is challenged in the event he testifies. This is true because of the particular evidentiary significance of coconspirator statements, which the Supreme Court explained in United States v. Inadi, 475 U.S. 387, 106 S.Ct. 1121, 89 L.Ed.2d 390 (1986):

Because they are made while the conspiracy is in progress, such statements provide evidence of the conspiracy’s context that cannot be replicated, even if the declarant testifies to the same matters in court.... Conspirators are likely to speak differently when talking to each other in furtherance of their illegal aims than when testifying on the witness stand. Even when the declarant takes the stand, his in-court testimony seldom will reproduce a significant portion of the evidentiary value of his statements during the course of the conspiracy.

Id. at 395, 106 S.Ct. at 1126. In this case, Wesley Mitchell and Jerry Mitchell did not appear as witnesses until after Agent High had testified as to statements made by the two men during the undercover investigation. High’s testimony as to these statements was admissible against appellants as Rule 801(d)(2)(E) co-conspirators’ statements and, therefore, was offered as significant substantive evidence of guilt, not as premature rehabilitation of subsequent witnesses as in Bolick. Consequently, we find no error in the admission of the statements.

Finally, appellants appear to claim that Agent High’s testimony about the statements of non-testifying co-defendants was somehow inadmissible. Unlike a confession or guilty plea of a co-defendant, however, the statements were plainly made “by a coeonspirator of a party during the course and in furtherance of the conspiracy,” and hence are admissible under Rule 801(d)(2)(E).

III.

Appellant Mitchell presented a defense of entrapment at trial, and now claims that the government did not make a sufficient showing of his predisposition to commit the crimes. It is settled law that “a valid entrapment defense has two related elements: government inducement of the crime, and a lack of predisposition on the part of the defendant to engage in the criminal conduct.” Mathews v. United States, 485 U.S. 58, 63, 108 S.Ct. 883, 886, 99 L.Ed.2d 54 (1988). The defendant has the initial burden of presenting more than a scintilla of evidence establishing these two elements, see United States v. Perl, 584 F.2d 1316, 1321 (4th Cir.1978), and if that burden is satisfied, “[t]he question of entrapment is generally one for the jury, rather than for the court.” Mathews, 485 U.S. at 63, 108 S.Ct. at 886. Even though the entrapment defense was considered by the jury in this case, Mitchell takes exception to the jury instructions because they did not state that the government had to prove that it was aware of Mitchell’s predisposition to engage in criminal conduct before it contacted him.

We find no error in the failure to make defendant’s proposed addition to the entrapment instruction. The focus of the predisposition inquiry must be on whether the defendant possessed the necessary state of mind to commit the crime when he *1258 was initially approached. See United States v. Hunt, 749 F.2d 1078, 1085 (4th Cir.1984). The government’s knowledge of this state of mind is irrelevant, and thus there is no requirement that the government prove that it had a reasonable suspicion of wrongdoing before conducting an undercover investigation. See, e.g., United States v. DeVore, 423 F.2d 1069, 1071 (4th Cir.1970); United States v. Luttrell, 923 F.2d 764 (9th Cir.1991) (en banc); United States v. Jenrette, 744 F.2d 817, 824 & n. 13 (D.C.Cir.1984); United States v. Gamble, 737 F.2d 853, 860 (10th Cir.1984); United States v. Jannotti, 673 F.2d 578, 608-09 (3d Cir.1982) (en banc).

Our review of the record also reveals that there was clear evidence of Mitchell’s predisposition introduced at trial. For example, the evidence showed that over seven months before High first contacted Joel Mitchell, High was told by Wesley Mitchell that Wesley could obtain cocaine from his cousin. When High indicated a desire to purchase cocaine, Wesley had High and Bear drive him to the home of Joel Mitchell, Wesley’s cousin, where Wesley obtained cocaine to resell to High. As another example of Mitchell’s predisposition, High testified that when he initially asked defendant Mitchell if he would be interested in purchasing cocaine or marijuana, Mitchell responded that he “might be interested in both.” Based upon this and other evidence, we have no difficulty concluding that there was ample basis for the jury to reject Mitchell’s use of the entrapment defense in this case.

IV.

Appellants next assert that the district court improperly refused to require the disclosure of the identity of the government informant who assisted Agent High in the undercover investigation. In Roviaro v. United States, 353 U.S. 53, 77 S.Ct. 623, 1 L.Ed.2d 639 (1957), the Supreme Court rejected a per se approach to the disclosure of an informant’s identity:

We believe that no fixed rule with respect to disclosure is justifiable. The problem is one that calls for balancing the public interest in protecting the flow of information against the individual’s right to prepare his defense. Whether a proper balance renders nondisclosure erroneous must depend on the particular circumstances of each case, taking into consideration the crimes charged, the possible defenses, the possible significance of the informer’s testimony, and other relevant factors.

Id. at 62.

Among those factors that must be taken into account in deciding whether to require disclosure is the scope of the informant’s role in the specific investigation. As this circuit has recognized, when the informant is an active participant in the transactions at issue instead of just a mere tipster, the failure to require disclosure of the informant’s identity is more likely to amount to error. See United States v. Price, 783 F.2d 1132, 1138-39 (4th Cir.1986); McL awhorn v. North Carolina, 484 F.2d 1, 4-5 (4th Cir.1973). This is particularly true in cases where the informant is the only witness other than the defendant who has first hand knowledge of critical portions of the criminal activity. See, e.g., Roviaro, 353 U.S. at 63-64, 77 S.Ct. at 629-30; Price, 783 F.2d at 1139. In this case, Bear was paid approximately $2000 per month by the DEA, and was involved in a number of DEA operations in addition to the instant one. Although there were a few instances where he took action by himself in this investigation, his main role was that of a supporting player to Agent High. Bear, nonetheless, was present to observe virtually all of the crucial events leading up to the arrest of appellants — a factor which should have been considered in the Roviaro balance had defendants moved for disclosure in a timely way.

Defendants, however, did not make the requisite showing to the district court of their need for the disclosure of Bear’s identity, and we therefore find no error in the district court’s failure to mandate disclosure. The burden is on the defendant to establish that the Roviaro criteria in a particular case counsel in favor of disclosure. See Rugendorf v. United States, *1259 376 U.S. 528, 534-35, 84 S.Ct. 825, 829-30, 11 L.Ed.2d 887 (1964); United States v. Johnson, 892 F.2d 707, 710 (8th Cir.1989). Indeed, this circuit has made clear that the onus is on the defendant to “come forward with something more than speculation as to the usefulness of such disclosure.” United States v. Smith, 780 F.2d 1102, 1108 (4th Cir.1985) (en banc); see also United States v. Martinez, 922 F.2d 914, 921 (1st Cir.1991); 2 Jack B. Weinstein & Margaret A. Berger, Weinstein’s Evidence ¶ 510[06] (1991) (“The defendant must explain to the court as precisely as possible what testimony he thinks the informer could give and how this testimony would be relevant to a material issue of guilt or innocence.”).

The record in this case reveals that appellants did not present the district court with any facts or evidence or any request for a hearing regarding their need to ascertain Bear’s identity. Their relative diffidence on this issue contrasts markedly with the assertiveness of defense counsel with respect to other evidentiary matters. Appellants claim that they filed a pretrial motion requesting that the government be ordered to reveal Bear’s identity, but it appears from the record that no such specific request was made. Instead, as part of a “Motion to Compel Disclosure of Existence and Substance of Promises of Immunity, Leniency or Preferential Treatment,” appellants included a generic request for impeachment information relating to any paid informants. At none of the three pretrial hearings did appellants raise the issue of disclosure despite being asked repeatedly by the district court if there were any matters that had not been discussed that required resolution before trial. The only point at which defense counsel made any specific request for Bear’s identity was when Mitchell’s counsel asked Agent High during cross examination what Bear’s name was. The government immediately objected on the ground that it was unnecessary for Bear to be identified since he would not be called by the government. The court sustained this objection, and defense counsel again gave no indication to the court why disclosure was necessary.

After all the evidence had been presented and the jury had been charged, appellants made an objection to the jury instructions related to the failure to disclose Bear’s identity. Apparently, appellants had requested that the jury be instructed that the government’s failure to call Bear should be taken as an inference that any testimony he would have provided would have been favorable to the defense. The court rejected this instruction, and in so doing, commented on the defense’s failure to raise the issue of disclosure earlier:

I think, also, in this case there could have been discovery instituted as to who Bear was and some compelling reason why it was absolutely necessary that he be here, and the Court could have heard all that in ample opportunity to'have considered that as a separate issue and gone into it.

In addition to these compelling observations by the district court, we note that the defendants even at that juncture failed to make the required showing of need for disclosure of Bear’s identity.

We are particularly unmoved by the appellants’ allegation of error on this point given their strategic use of Bear’s absence at trial. Joel Mitchell testified at trial that Bear repeatedly pressured him to make the drug purchase from High, and Mitchell even claimed that Bear provided the money to Mitchell for the March 15 purchase. Yet, despite this crucial role that Bear allegedly played in entrapping Mitchell, Mitchell apparently did not ask the court to compel Bear to testify. Instead, in closing arguments, counsel for Mitchell made Bear’s absence a primary basis for arguing that reasonable doubt existed, and Blevins’ attorney used Bear’s absence in a similar manner. When the appellants objected to the failure to give the requested instruction to the jury regarding Bear’s absence, the district court indicated that it was aware of just how much the defendants stood to benefit from Bear’s absence:

I don’t know of a thing, really, that Bear could have added except he could have denied the story of Mr. Mitchell, but perhaps Mr. Mitchell should be very thank *1260 ful that he’s not here so that his story could be uncontradicted.

The defendants’ decision to forego any attempt to convince the trial court that disclosure was necessary in this case appears to have been a strategic gamble, and “risky gambling tactics such as this are usually binding on the gambler.” Gonzalez v. Volvo of Am. Corp., 752 F.2d 295, 298 (7th Cir.1985). The decision whether to require disclosure of an informant’s identity is typically within the discretion of the trial court. See Smith, 780 F.2d at 1108; United States v. Gonzalez, 555 F.2d 308, 314 (2d Cir.1977). Given that appellants never made a proper showing of the need for disclosure and that they derived substantial benefit from and made significant use of the informant’s absence, we find no error in the failure to require disclosure of Bear’s identity in this case.

Appellants also accuse the government of improperly attempting to bolster the credibility of Bear by questioning Agent High as to High’s previous involvement with Bear. Appellants’ claim that High testified to Bear’s “long standing cooperation with the DEA and his considerable success” is incorrect. It was counsel for Mitchell and Johnson who initially questioned High as to his previous association with Bear, and the government simply asked on redirect examination how many different investigations High and Bear had worked on together. There was no discussion of any kind as to the success of these past operations, and appellants’ suggestion that the government was somehow trying to tempt the jury to accept the opinions of other juries in other cases is badly misplaced.

An assignment of error that does have merit is appellants’ claim that the district court should not have allowed into evidence testimony that the six non-testifying co-defendants 2 had pleaded guilty. Courts have generally agreed that evidence of a non-testifying co-defendant’s guilty plea should not be put before the jury. See, e.g., United States v. Leach, 918 F.2d 464, 467 (5th Cir.1990); United States v. De La Vega, 913 F.2d 861, 866 (11th Cir.1990). The reason for this prohibition is twofold. First, by not having the opportunity to cross-examine the co-defendant who entered the guilty plea, the defendant on trial is unable to probe the motivations for entry of the plea. This significantly undercuts the defendant’s right to have a jury’s verdict based only upon evidence that is presented in open court and is thereby subject to scrutiny by the defendant. See United States v. Griffin, 778 F.2d 707, 711 (11th Cir.1985) (citing Turner v. Louisiana, 379 U.S. 466, 472-73, 85 S.Ct. 546, 549-50, 13 L.Ed.2d 424 (1965)). Second, introduction of such guilty pleas raises the concern that a defendant might be convicted based upon the disposition of the charges against the co-defendants, rather than upon an individual assessment of the remaining defendant’s personal culpability. See Griffin, 778 F.2d at 711.

Because of these two concerns, any mention of guilty pleas of nontestifying co-defendants should be avoided at trial wherever possible. When references to the resolution of charges against co-defendants must be made, it is “preferable to tell the jury only that the case[s] against the co-defendants had been disposed of previously and that they had been eliminated from the present trial.” United States v. Curry, 512 F.2d 1299, 1303 (4th Cir.1975). If for whatever reason the jury does learn that co-defendants have pleaded guilty, the court upon request should issue a limiting instruction to jurors stating that the evidence of such guilty pleas is not to be taken as substantive evidence of guilt of the remaining defendants. See id. 3

*1261 During the course of this seven-day trial, the guilty pleas of the six non-testifying co-defendants were briefly but improperly discussed on three occasions. 4 First, during Agent High’s testimony on the second day of the trial, High identified a series of exhibits related to drug purchases he had made from the six non-testifying eo-defen-dants, as well as two testifying co-defendants. The government then asked High what had happened to the eight individuals, and he responded that they had all pleaded guilty to certain counts within the indictment. Second, on the third day of the trial when a drug analyst was called to testify to the composition of the drugs that High had purchased from the eight co-defendants, the judge asked defense counsel if they were willing to stipulate as to the chemical composition of the drugs, and the judge remarked in passing that the people who sold these drugs had already pleaded guilty. Finally, in the government’s closing argument, the prosecutor responded to numerous arguments made by defense counsel. To counter the defense charge that the government had wasted money by purchasing drugs during the undercover operation, the government indicated that a relatively small expenditure had resulted in eight individuals pleading guilty and three additional defendants being tried. The prosecutor also noted in the closing argument that in order to explain how the investigation developed, evidence had been presented relating to “defendants that pled guilty.” While these references to the guilty pleas were in no way a significant part of the case against appellants, it was still error for any mention of the pleas of the non-testifying co-defendants to have been made. The question we thus must address is whether such error mandates reversal of the convictions.

Appellate courts long ago rejected the notion that reversal is necessary for any error committed by a trial court. Our judicial system is populated by fallible human beings, and some error is virtually certain to creep into even the most carefully tried case. The ultimate aim of the system, therefore, is not unattainable perfection, but rather “fair and correct judgments.” Rose v. Clark, 478 U.S. 570, 579, 106 S.Ct. 3101, 3106, 92 L.Ed.2d 460 (1986). When a court blindly orders reversal of a judgment for an error without making any attempt to assess the impact of the error on the outcome of the trial, the court “encourages litigants to abuse the judicial process and bestirs the public to ridicule it.” Roger J. Traynor, The Riddle of Harmless Error 50 (1970). As a practical response to the realities of the trial process, therefore, appel *1262 late courts have developed a series of doctrines for analyzing whether error in various types of cases was harmless. The fundamental premise of this sort of analysis is that “[a] defendant is entitled to a fair trial but not a perfect one.” Lutwak v. United States, 344 U.S. 604, 619, 73 S.Ct. 481, 490, 97 L.Ed. 593 (1953); accord Delaware v. Van Arsdall, 475 U.S. 673, 681, 106 S.Ct. 1431, 1436, 89 L.Ed.2d 674 (1986).

This harmless error analysis is not inapplicable to drug conspiracy cases. While courts must always be vigilant in a conspiracy case in their “use of every safeguard to individualize each defendant in his relation to the mass,” Kotteakos v. United States, 328 U.S. 750, 773, 66 S.Ct. 1239, 1252, 90 L.Ed. 1557 (1946), judges must also remember that “[w]hen many conspire, they invite mass trial by their conduct.” Id. Assessment of whether an error was harmless is certainly not an easy task in a complex conspiracy case, but to fail to apply the analysis is to risk systemic paralysis.

The question remains, of course, whether harmless error analysis should be applied to the particular type of error in this case. We believe that the error in introducing the guilty pleas of non-testifying co-defendants is of constitutional dimension. We look, therefore, to the Supreme Court’s recent clarification of the standard to be used in deciding which constitutional errors may be deemed harmless. See Arizona v. Fulminante, — U.S. -, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991). Whether harmless error analysis applies depends upon whether the error is best classified as a trial error or a structural error. A trial error is one “which occurred during the presentation of the case to the jury, and which may therefore be quantitatively assessed in the context of other evidence presented in order to determine whether its admission was harmless beyond a reasonable doubt.” Id. Ill S.Ct. at 1264. These trial errors constitute the overwhelming majority of constitutional errors, but there are a limited number of structural errors for which harmless error analysis would be inappropriate. Structural errors are ones that affect “[t]he entire conduct of the trial from beginning to end,” such that any attempt by a reviewing court to isolate the impact of the error would be fruitless. Id. at 1265. Examples of this latter type of error include the deprivation of the right to counsel and the presence at trial of a biased judge. Id.

Admission of the guilty pleas of non-testifying co-defendants provides a classic example of a trial error to which harmless error analysis may be applied. The erroneous discussion of the guilty pleas came at discrete points during the presentation of the case to the jury, and the impact of the error can be evaluated in light of the other evidence which was properly admitted. The error simply cannot be classified as a structural one that infected the entire trial process.

Our decision to apply harmless error analysis in this case is further bolstered by the Supreme Court’s application of harmless error doctrine to an analogous area of law. In Bruton v. United States, 391 U.S. 123, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968), the Supreme Court held that evidence of a non-testifying co-defendant’s confession that inculpates another defendant could not be introduced at a joint trial without violation of the other defendant’s constitutional rights. Id. at 126, 88 S.Ct. at 1622-23. In subsequent cases, however, the Court refused to overturn convictions because the Bruton error in the individual cases was deemed to be harmless. See Schneble v. Florida, 405 U.S. 427, 430, 92 S.Ct. 1056, 1058-59, 31 L.Ed.2d 340 (1972); Harrington v. California, 395 U.S. 250, 254, 89 S.Ct. 1726, 1728-29, 23 L.Ed.2d 284 (1969). If erroneous admission of a co-defendant’s confession can be harmless, the harmless error doctrine certainly must be applicable to errors involving a co-defendant’s guilty plea, given that a guilty plea is in essence a “barebones” confession.

In deciding whether a constitutional error was harmless, a reviewing court must be satisfied that the error was harmless beyond a reasonable doubt. See, e.g., Ful- *1263 minante, 111 S.Ct. at 1264; Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967). While the concept of harmlessness protects the public interest in convictions where it is clear that the defendant committed the criminal offense, the reasonable doubt standard serves the no less important interest in safeguarding our defining freedoms. Here we must ask whether it is clear beyond a reasonable doubt that the jury would have returned verdicts of guilty against appellants even if the evidence concerning the co-defendants’ guilty pleas had not been introduced. See United States v. Hasting, 461 U.S. 499, 510-11, 103 S.Ct. 1974, 1981-82, 76 L.Ed.2d 96 (1983). As discussed above, the decision of this question requires a quantitative assessment of the likely impact of the error measured against the other evidence presented at trial. Fulminante, 111 S.Ct. at 1264; see also Schneble, 405 U.S. at 431, 92 S.Ct. at 1059; Harrington, 395 U.S. at 254, 89 S.Ct. at 1728-29 (both cases focusing on the overwhelming evidence of guilt in finding errors harmless). Based upon our examination of the record, we are convinced that the error here was harmless beyond a reasonable doubt. We are compelled to reach this conclusion by the overwhelming nature of the evidence against these three appellants. Careful review of that evidence will help to assure that defendants’ essential right to a fair trial has not been abridged.

Much of the evidence of Joel Mitchell’s guilt is uncontroverted. During his testimony at trial, Mitchell conceded that after a series of telephone calls and face-to-face meetings with High, Mitchell agreed to purchase three kilos of cocaine and 35 pounds of marijuana. Mitchell also admitted that he, High, Bear, and Blevins toured his property on March 2 in search of potential drop sites for the drugs and that he participated in the practice air drop on March 13. He also did not contest that he was present at the agreed site for the drug delivery on March 15 with half of the agreed purchase price as well as a handgun in his possession. Although he claimed not to have noticed the siren or flashing lights used by the arrest team, he did not challenge the testimony of at least five of the arresting officers that he tried to escape when the officers first arrived at the scene of the drug bust.

In his testimony, Mitchell claimed that he had never before been involved in the sale of cocaine or marijuana and that he was not purchasing the drugs from High in order to sell them. Instead, Mitchell testified that he was purchasing the drugs for Bear and was only going to hold them overnight, for which he would be paid $30,-000. Mitchell also stated that Bear initiated the scheme, that he acted at Bear’s direction throughout the events at issue, and that Bear provided him the money to make the deal.

It is hardly surprising, however, that the jury looked askance at Mitchell’s tale. Recorded tapes of Mitchell’s conversation with High and Bear on the day of the practice air drop include numerous statements that undercut Mitchell’s claims that he had not previously been a drug dealer and that his interest in this transaction was not that of a drug distributor. During the tape recorded conversation, Mitchell stated that his girlfriend was aware of the fact that he was “dealing” drugs. Mitchell also stated that three kilos of cocaine would not last him long and that if the quality of the cocaine was adequate, he would require five, six, or seven kilos per month in future deliveries. At another point in the tape recording, Mitchell mentioned that he had a partner who might be present on the day of the drug delivery. Despite Mitchell’s denial during his testimony that he ever said that his partner would get half of the drugs, the tape recording contains a statement by Mitchell that his partner would be “takin’ half of it.” These statements regarding a partner, as well as the others related by Agent High in which Mitchell actually provided a description of his partner, undercut Mitchell’s claim that he was acting solely at Bear’s behest.

Mitchell’s testimony is also contradicted by probative evidence on other crucial points. For example, Mitchell testified that Bear delivered the money for the drug purchase to his house during the early eve *1264 ning of March 14, but Agent High testified that Bear was with him and several other investigators in Roanoke, Virginia at that time. Mitchell’s claim that he did not sell drugs to Wesley Mitchell on June 14, 1989, is similarly contradicted. Wesley Mitchell, Joel's cousin, testified that when High sought to purchase cocaine from Wesley on that date, Wesley called Joel, who told him to come to his house to obtain the necessary cocaine. Wesley further testified that after he was driven to Joel’s house, Wesley went inside by himself, where he paid Joel $400 in exchange for six grams of cocaine which were handed to Wesley by Joel’s brother Paul. As corroboration of Wesley’s account, High testified that he did, in fact, accompany Wesley to Joel’s residence, that Wesley told him that his cousin could supply the desired cocaine, and that Wesley returned from Joel’s house with six grams of cocaine.

While we could recite many more pieces of evidence that inculpate Joel Mitchell, we need not engage in such an endeavor. After reviewing the extensive evidence that was presented at trial, we consider the evidence of guilt against Mitchell to be overwhelming.

The evidence presented against Ervin Blevins was similarly compelling. Blevins admits that he was present for the March 2 meeting between High and Mitchell; that he went along with Mitchell, High, and Bear on the March 2 tour of the Mitchell property; that he saw the large amount of cash in Mitchell’s possession on the day of the purchase; and that he was with Mitchell and was in possession of a handgun when the arrest was made. Blevins testified, however, that he was completely unaware that any drug purchase was being contemplated. The evidence does not support Blevins’ claim of ignorance.

Agent High testified that Blevins overheard the discussions as to price and quantity that took place at Mitchell’s residence on March 2, and High also stated that Blevins was present during extensive discussions of the best locales for an air drop. High also recounted that Mitchell asked Blevins on March 2 if Blevins was “in on this thing with me,” and Blevins responded that he was. High further testified that when Mitchell asked Blevins to assure High that Mitchell was good for the money, Blevins told High that “[i]f he tells you he’s good for the money, you can believe him he’s good for the money.” High also stated when High asked whether the police were likely to make a raid, Blevins responded, “If the police come, we’re ready for them.”

In addition to these incriminating statements, there was also evidence from other sources against Blevins. The tape recording of the March 13 practice air drop includes a discussion between Mitchell, High, and Bear as to who would be present on the day of the delivery and what procedures would be followed. The tape reveals that Mitchell stated that Blevins would be present with him at the time of the drop and that Mitchell planned to have Blevins wait with High and Bear while Mitchell retrieved the drugs. Blevins was indeed with Mitchell at the site of the air drop on March 15. Two of the arresting officers testified that when they arrived in their vehicles to arrest Mitchell and Blevins, Blevins ran to Mitchell’s truck and reached through the window of the truck rather than for the door handle. After the arrest had been completed, the officers found Blevins’ handgun sitting on the seat of the truck. These examples again only highlight the strong evidence of Blevins’ guilt.

The evidence against Dewey Johnson presented by the government was equally overwhelming. Agent High testified that Mitchell described his partner as a man with a big, brown, bushy mustache who had been arrested for an offense involving a large quantity of marijuana and who at the time of that arrest had used or possessed a handgun. Evidence at trial showed that Dewey Johnson matched that description. High also testified that he saw Dewey Johnson leaving Joel Mitchell’s house just as High arrived on the day of the arrest and that when High went inside, Mitchell told him that his partner had just left. According to High, Mitchell also stated that his partner would be waiting with half of the money at Mitchell’s vacant *1265 store. The officers who arrested Johnson testified that Johnson was found by himself in the vacant store at the time of the planned drug delivery. As discussed above, the officers also found $6400 on Johnson’s person and $35,500 hidden in the store, cash totalling just over half the purchase price. One of the officers testified that Johnson told him that he was at the store to level a pool table, yet a search of the area revealed none of the equipment required to do that job. Taken together, these pieces of evidence provide a clear picture of Johnson’s guilt.

After surveying the evidence against the three appellants, we are convinced beyond a reasonable doubt that the jury’s verdict would not have been different had the minimal discussion of the non-testifying eo-de-fendants’ guilty pleas not taken place. We therefore do not deem the error committed at trial to warrant reversal of these convictions.

VI.

Appellants allege a long list of errors committed by the district court. This string of contentions includes claims that there was inadequate evidence to support the verdicts, that joinder of Johnson was improper, that multiple conspiracies were proved although only a single conspiracy was charged, that the jury was improperly instructed on the weapons possession charge, that evidence of other crimes committed by Blevins was introduced without adequate warning, and so on. This extended litany of allegations does nothing to shake our conviction that justice was done in this case. Our examination of the record leads us to conclude that none of the remaining assignments of error have merit. The convictions of appellants are therefore

AFFIRMED.

. Johnson also contends that the trial court erred in admitting testimony concerning his 1979 arrest and subsequent conviction for marijuana and weapons charges. This evidence was plainly admissible under Federal Rule of Evidence 404(b) to prove Johnson’s identity, and the district court properly indicated that the scope of testimony on the subject should be *1257 confined to the facts required to corroborate Mitchell’s description of his partner.

. The six “co-defendants" were not tried with appellants. The six, however, were indicted along with appellants and pleaded guilty prior to appellants’ trial. Given that other courts have referred to similarly situated individuals as "co-defendants,” we adhere to that convention.

. As to the pleas of the two testifying co-defendants, we are less inclined to find any error in *1261 the discussion of their pleas given that the appellants had the opportunity to cross-examine them. Moreover, the government was justified in eliciting testimony as to their pleas in anticipation of impeachment of the witnesses on that basis. If there was any error, however, it was rendered harmless by the district court’s instruction to the jury at the end of the trial that they were not to consider the guilty pleas of the testifying codefendants as substantive evidence against appellants. Because this limiting instruction for whatever reason did not refer as well to the pleas of the nontestifying co-defendants, we have not relied upon it in tbe analysis of harmless error in this section.

. In addressing this issue, we assume that timely and proper objections were made to testimony with respect to the guilty pleas. We note, however, that it requires a charitable view of the record to deem appellants' objections on this point sufficient. At the time Agent High testified that the eight codefendants had pleaded guilty, counsel for Johnson objected only on the basis that High had misstated the exact charges to which the individuals had pleaded guilty. There was no contemporaneous objection that any testimony with respect to the pleas was improper.

Appellants’ claim that their motion in limine seeking exclusion of plea agreements of the co-defendants covered any mention of guilty pleas is also somewhat strained. Appellants' stated rationale in making that motion was not to preclude any mention of a guilty plea but to keep out elements of a plea agreement that might vouch for a witness' credibility — such as plea agreement provisions requiring that the witness tell the truth.

If these objections were inadequate, appellants would be left to argue that introduction of evidence concerning the guilty pleas was plain error. The standard for reversal based on plain error is more difficult for appellants to satisfy than that for harmless error. See United States v. Young, 470 U.S. 1, 17 n. 14, 105 S.Ct. 1038, 1047 n. 14, 84 L.Ed.2d 1 (1985).