In September, 2001, the defendant, Aaron Kincaid, was convicted by a Superior Court jury on four indictments charging aggravated rape.1 G. L. c. 265, § 22 (a). Following a postverdict inquiry of the jurors,2 the trial judge determined that *382the jury had been exposed to an extraneous matter (the flight of Kincaid’s coventurer), and ordered a new trial. The Commonwealth appealed and the Appeals Court reversed. Commonwealth v. Kincaid, 61 Mass. App. Ct. 657 (2004). We granted the defendant’s application for further appellate review, and now affirm the trial judge’s order.
1. Facts. We summarize the trial evidence only to the extent relevant to the postverdict inquiry.
a. The trial. The Commonwealth’s theory was that the defendant and Richard Lampron, as joint venturers, raped Lam-pron’s former girl friend, whom we call Gail. The defense theory was consent. According to Gail’s trial testimony, on September 23 and 24, 1998, while she was stupefied, the defendant and Lampron engaged in sexual acts with her against her will. In addition, a “barrette”3 was inserted into her vagina. There was testimony that the episode was recorded on videotape.
Two friends of the defendant and Lampron, who had viewed portions of the videotape, stated that they believed that Gail appeared nonresponsive. According to one of them, the videotape depicted Lampron inserting the “barrette” into her vagina.
Following the incident, in the presence of Lampron, the defendant described to other acquaintances what “we did to [Gail],” including sticking “a hair clip or a barrette” inside her and not being able to get it out; that Gail was “unconscious”; that they took a “muscle relaxer” from her purse; and that Lampron then cautioned the defendant not to reveal any more.
During the police investigation of the crime, the defendant gave the police a detailed written description of the incident in which he denied having sexual contact with Gail, and maintained that he had witnessed Lampron engaging in consensual sexual activity with her. Deoxyribonucleic acid (DNA) testing showed, however, that DNA found on the “barrette” was consistent with having come from the defendant, Gail, and a second male.
At trial, the defendant, faced with the DNA evidence, changed his story and claimed that the intercourse was consensual. His *383contention was supported by Erica Dewey, his girl friend at the time of trial. She testified that she had viewed the videotape, and that it indicated Gail was aware of what was happening, actively participated in the sexual activity with the defendant and Lampron, and did not appear incapacitated or dragged.
The defendant testified, stating that he had hidden the videotape of the incident in the trunk of his car because he believed Lampron would try to take it from him. The defendant also said that, after he told Lampron that he was being investigated and was going to the district attorney’s office, Lampron did take the videotape. The videotape was not produced at trial and its disappearance and contents were the subject of much trial testimony.
At the time of trial, September, 2001, Lampron was a fugitive, his whereabouts unknown.4 The judge and counsel took pains to keep this information from the jury and there was no reference to Lampron’s flight or his status as a fugitive. To this end, the judge instructed that “for various legal reasons it is not uncommon that persons charged as coventurers are tried separately, and you are not to speculate as to why that is the case here. It is irrelevant to your proper deliberations.”
b. The postverdict inquiry. In May, 2002, the defendant moved for a postverdict inquiry claiming that the jury had been exposed to a significant extraneous fact: Lampron’s flight. The defendant’s attorney filed an affidavit stating that one deliberating juror, juror A, claimed that the jurors might have some information about Lampron’s flight that they had not learned at trial, and that the information had influenced her decision. The judge conducted hearings in July and August of 2002, during which he inquired of all the jurors to determine whether they had learned anything of Lampron’s absence.
The standard of review of findings made in a postverdict inquiry is clear error. “There is no reason to give a judge’s finding of fact less weight in a postverdict context than we ordinarily would, i.e., we accept his finding unless clearly *384erroneous.” Commonwealth v. Ciminera, 11 Mass. App. Ct. 101, 109, S.C., 384 Mass. 807 (1981). “[A] finding of fact by the trial judge will not be deemed ‘clearly erroneous’ unless the reviewing court on the entire evidence is left with the firm conviction that a mistake has been committed.” Commonwealth v. Tavares, 385 Mass. 140, 156, cert. denied, 457 U.S. 1137 (1982), quoting New England Canteen Serv., Inc. v. Ashley, 372 Mass. 671, 675 (1977).
We summarize the testimony at the hearing on the motion for a postverdict inquiry. Juror A (who initially had reported her concerns) stated that another juror, in discussing his view of the defendant’s guilt, “alluded to the fact that the other person named had taken off,” and that she had not known that. She stated that a third juror said, “We’re not supposed to know that.” Juror A indicated that this information affected her own view of the defendant’s guilt. She also stated that the juror who initially mentioned Lamprori’s flight spoke with a distinctive accent.
Juror B, the juror with the distinctive accent, testified that there had been “mention” of Lampron’s absence, but denied that he had raised the subject. He could not recall which juror had mentioned the topic first and also believed that there was trial testimony about Lampron’s flight.
In addition to jurors A and B, five others remembered discussion during deliberations of Lampron’s absence. Juror C stated that the jury were curious about Lampron’s absence, and that she thought a juror with a distinctive accent stated, “I think I’ve heard that he’s no longer around here, he took off; something to that effect,” and that either she or another juror objected to this statement. Juror C also stated that the jurors agreed that it was not information introduced to them and that they should not consider it in their decision-making process. Juror D believed that there was probably discussion about Lam-pron’s flight, but thought that the jurors had learned of it from evidence presented at trial, perhaps from the testimony regarding the missing videotape. Juror E’s recollection was that the jury assumed that Lampron “was on the run,” but did not remember how the issue came up. Juror F recalled that the jurors had mentioned Lampron’s flight but could not say who *385stated that Lampron had fled and thought there might have been evidence about the subject during trial. Juror G inferred that she must have learned of Lampron’s flight through some comment by another juror, but could not remember specifically how she had come to that conclusion. Five jurors did not recall any mention of Lampron’s having “taken off,” “being on the run,” or any similar explanation for his absence.
The judge made careful and detailed findings and included bis assessment of the credibility of the jurors’ testimony at the hearing. The judge found that “at one point a juror raised the issue of Lampron’s absence; it may have been by a comment or a question,” that “[ajfter Lampron’s name came up, another juror stated, in substance, that Lampron was on the run, that is, that he had fled.” The judge further found that one juror, “in explaining his thinking[,j alluded to the fact that [Lampron] had taken off.” The judge found it probable that juror B made the statement that Lampron was “on the run.” He did so on the basis that the testimony of jurors A and C was credible, and because he did not credit juror B’s denial of making such a statement. He doubted juror B’s credibility in part because juror B “noticeably hesitated” when asked “the relevant question.” The judge concluded that the juror’s hesitation stemmed from his “recognition that he said something during deliberations that he should not have.”
The judge concluded that “it is more probable that there had been an explicit, fact-like reference to Lampron’s flight rather than the jury having reasoned, opined, or assumed its way to that conclusion.” The judge determined that the defendant had shown, by a preponderance of the evidence, that the jury were exposed to extraneous matter during their deliberations. He further ruled that the Commonwealth had not proved beyond a reasonable doubt that the defendant was not prejudiced by the extraneous matter. Specifically, the judge determined that, aside from the “ordinary inference” drawn from a coventurer’s flight, the jury would make the further inference (based on Kincaid’s testimony that Lampron had stolen the videotape) that Lampron had fled with the tape because it showed that the sexual activity was nonconsensual. In reaching his decision, the judge declined to consider juror A’s testimony that learning of Lampron’s *386flight affected her view of the case. The judge agreed with the defendant’s contention in his posthearing memorandum that a new trial was warranted.
The Commonwealth appealed to the Appeals Court on the ground that (1) the judge’s finding that the jury had been exposed to an extraneous matter was clearly erroneous and that on the evidence the jury permissibly could infer that Lampron had fled; and (2) even if the jury had been exposed to extraneous matter, the Commonwealth had proved beyond a reasonable doubt that the defendant was not prejudiced. The Appeals Court agreed with the Commonwealth on both grounds. See Commonwealth v. Kincaid, 61 Mass. App. Ct. 657, 661-666 (2004).
2. Discussion. When there is a claim of extraneous influence on a jury, a two-step procedure is to be followed. First, the defendant “bears the burden of demonstrating that the jury were in fact exposed to the extraneous matter. To meet this burden he may rely on juror testimony.” Commonwealth v. Fidler, 377 Mass. 192, 201 (1979). If the defendant meets this burden and the judge finds that extraneous matter came to the attention of the jury, “the burden then shifts to the Commonwealth to show beyond a reasonable doubt that [the defendant] was not prejudiced by the extraneous matter.” Id.
The judge here followed the Fidler procedure and concluded, with respect to the first step, that the defendant had satisfied his burden of proof by showing, by a preponderance of the evidence, that the jury were exposed to extraneous matter during their deliberations. Although we have not previously stated explicitly the level of proof required for the defendant to meet his burden to show that the jury were exposed to extraneous matter, we today hold that a defendant’s burden is to prove exposure to extraneous matter by a preponderance of the evidence.5 We adopt this standard because once there is a suggestion that the process has been tainted, we should not make it excessively difficult for the defendant to establish the existence *387of the taint. The preponderance standard is sufficient, particularly in light of the fact that it applies only to the first part of a two-step procedure. Other cases have imposed a similar burden on the defendant when there is a suggestion of improper influence on jurors. See, e.g., Commonwealth v. Amirault, 399 Mass. 617, 626 (1987) (defendant bears burden of showing juror is not impartial by preponderance of evidence); Commonwealth v. Laguer, 36 Mass. App. Ct. 310, 314 (1994) (defendant “had the burden of demonstrating by a preponderance of the evidence that the jury were exposed to ethnic or racial bias”). The judge applied the correct burden of proof to the first stage of the procedure.
Applying the standard of review previously discussed, we consider whether the judge properly concluded that the jury were exposed to extraneous matter. While this is a close question, we must defer to the findings of the judge unless those findings are clearly erroneous. The dispute is essentially whether juror B was repeating something that he learned from an extraneous source or was speculating as to the reason for Lam-pron’s absence. While it is possible that the jury could have concluded that Lampron had fled because of evidence adduced at trial, there was also evidence that this matter came from an outside source.
The resolution of this issue is a question of fact for the judge. The finding that juror B gleaned this information from an extraneous source is warranted by the fact that juror C testified that juror B said, “I think I’ve heard" (emphasis supplied). From this statement, the judge could draw a reasonable inference that “I think I’ve heard” does not mean “I am speculating” or “I am drawing my own conclusion,” but rather, “I think I was told.” The judge credited the testimony of the reporting witness (juror C) as he was entitled to do, and drew a reasonable inference from the reported statement that juror B received the information from a third party and was not merely speculating. That inference, while by no means compelled, was certainly permissible.
In addition, the judge relied on the recollections of other *388jurors that the juror who made the statement said that Lampron was “on the run” or “had fled.” The judge was entitled to conclude, as he did, that these words suggest a statement of fact, rather than an assumption. The judge also gave weight to the fact that juror B “noticeably hesitated” when asked if such a statement had been made, and concluded it was due to juror B’s having “said something during deliberations that he should not have.” As the judge is “in the best position to judge the weight and credibility of the evidence,” Commonwealth v. Tavares, 385 Mass. 140, 156 (1982), quoting New England Canteen Serv., Inc. v. Ashley, 372 Mass. 671, 675 (1997), we defer to his conclusions regarding the credibility of the various jurors’ testimony. See Commonwealth v. Lazarovich, 410 Mass. 466, 478 (1991) (“It was for the judge to evaluate the credibility of the witnesses who testified at the hearing”).
We note also that all parties made great effort during trial not to mention the reason for Lampron’s absence. The judge instructed the jury not to speculate on the reason that Lampron was not being tried with the defendant, telling them that separate trials of alleged coventurers were “not uncommon.” Taken together, the judge’s admonition and the parties’ forbearance reduce the likelihood that something occurred within the confines of the trial that would have invited the jury to speculate on the possibility that Lampron had fled or caused them to reach such a conclusion. By comparison, the defendant demonstrated that there had been considerable publicity concerning the case, including specific reporting in the media of Lampron’s status as a fugitive. Based on the record of the postverdict inquiry, we cannot say that the judge clearly erred in concluding that the source of the information was probably extraneous.
We do not find persuasive the Commonwealth’s contention that the lack of direct evidence of the source of juror B’s knowledge impairs a conclusion that extraneous matter was introduced to the jury. Our cases have not required identification of the source of a juror’s knowledge of the extraneous matter. It is only necessary that there be adequate evidence that the knowledge did not come from the evidence at trial. See Commonwealth v. Hunt, 392 Mass. 28, 39-40 (1984) (after *389verdict reached juror stated he knew defendant had criminal record); Commonwealth v. Fidler, 377 Mass. 192, 199-200 (1979) (finding extraneous influence where juror stated someone shot at defendant previous month). See also Jeffries v. Wood, 114 F.3d 1484, 1488 (9th Cir. 1997) (juror informed other jurors that defendant was convicted armed robber).
The case of Commonwealth v. Schoen, 24 Mass. App. Ct. 731, 733-734 (1987), relied on by the Commonwealth, is not to the contrary. In that case, the jury, while still deliberating, asked a question that suggested that they knew or had guessed that a person involved in the crime had pleaded guilty to a lesser included offense. Id. at 731-732. The judge forcefully instructed the jury not to speculate on what had happened to the other person, probed the jurors for extraneous influences, and found that there were none. Id. at 732-733. The Appeals Court concluded in that case that the judge had not erred in denying the defendants’ motion for a mistrial. Id. at 734. In the instant case, on the basis of permissible findings, the judge reached the opposite conclusion. Each time, the appellate court defers to the judge’s role in assessing juror credibility. Id. at 733-734.
We next consider whether the judge erred in the second step of the process, i.e., in determining that the Commonwealth had not met its burden of proving beyond a reasonable doubt that Kincaid was not prejudiced by the extraneous matter. In this assessment, the judge must focus on the probable effect of the extraneous fact on a “hypothetical average jury.” Commonwealth v. Fidler, supra at 201. “The judge may consider such matters as: (1) the fact that the improper remark prompted an immediate reprimand from another juror ... (2) whether there was overwhelming evidence of guilt. . . and (3) whether the extraneous matter produced such a high probability of prejudice that error must be inferred” (citations omitted). Id. at 201 n.8. The judge concluded that the first two factors weighed in favor of the Commonwealth. Juror B’s remark met with a prompt reprimand from juror C, and the Commonwealth “presented a strong, compelling case.” The judge decided, however, that the extraneous matter created a high probability of prejudice. He believed that the jury likely would reason “that Lampron fled because he was guilty, he was guilty because the *390intercourse was non-consensual, and the videotape must therefore reflect that fact.” Thus, knowledge of Lampron’s flight would corroborate the Commonwealth’s testimony concerning the videotape and undermine that of the defendant and his girl friend.
The judge’s conclusion, that Lampron’s flight in these circumstances created a high probability of prejudice, is not clearly erroneous. As the judge recognized, the flight of a co-venturer suggests his guilt, and suggests further that his partner is guilty as well. The likely assumption by the jury, that Lam-pron perceived that he and Kincaid were guilty, would create prejudice against Kincaid. Added to that is the significance of the missing videotape. That Lampron fled with the videotape (as suggested by Kincaid’s testimony that he believed Lampron had taken the videotape) indicated that the tape showed that the sexual activity was nonconsensual. This eviscerates Kincaid’s defense of consent and it does so based on totally extraneous information.
While not an inescapable conclusion that a hypothetical average jury would follow this logic, such a finding was warranted. The judge could conclude that the introduction of the extraneous matter would destroy the defense’s credibility. Because the judge permissibly found that there was an extraneous influence, and determined that the Commonwealth had not satisfied its burden to show that Kincaid was not prejudiced by the extraneous matter, a new trial must be granted. Having made the two foundational findings, the judge retains no discretion to deny a new trial. See Commonwealth v. Cuffie, 414 Mass. 632, 638 (1993).
The Commonwealth attempts to draw support for its position that the defendant was not prejudiced from cases in which we have stated that evidence of a codefendant’s flight does not require severance of joint trials under the rule established in Bruton v. United States, 391 U.S. 123 (1968). These cases do not hold that a jury’s knowledge of a codefendant’s flight is not prejudicial. Rather, the cases stand for the proposition that separate trials for each defendant are not mandated where the degree of prejudice is not so severe and there are strong, timely limiting instructions. See Commonwealth v. Pontes, 402 Mass. *391311, 314-315 (1988); Commonwealth v. McGrath, 358 Mass. 314, 321 (1970); Commonwealth v. Steven, 29 Mass. App. Ct. 978, 979 (1990); Commonwealth v. Clark, 5 Mass. App. Ct. 673, 676-677 (1977). The amount of prejudice created by a jury’s knowledge of a codefendant’s flight is attenuated enough that, when combined with proper limiting instructions, separate trials may not be constitutionally required. However, the amount of prejudice from knowledge of a codefendant’s flight is sufficient to justify a judge’s finding of prejudice, at least where, as here, the judge has determined that the peculiar circumstances of the case (including the absence of any curative instruction about flight) would lead a hypothetical average jury to draw an inference more damaging to the defendant than evidence of flight alone.
We mention one other matter. A judge may not inquire into the deliberative process of the jury. See Commonwealth v. Fidler, supra at 198 (“Our decision does not permit evidence concerning the subjective mental processes of jurors, such as the reasons for their decisions”); Commonwealth v. Royster, 15 Mass. App. Ct. 970 (1983); Commonwealth v. Scanlan, 9 Mass. App. Ct. 173, 184 (1980) (“we would embark on a slippery slope indeed if we began to monitor and evaluate the internal procedures of the jury”). We recognize the difficulty of conducting the Fidler inquiry without touching on forbidden ground.
Here, the judge attempted to confine his questioning to learning what was said in order to ascertain whether there was an extraneous influence during the jury’s deliberations. It may happen, as it did here, that a juror responds to a permissible question with an answer that inappropriately reveals aspects of the deliberations. Giving cautionary instructions to each juror at the outset of the inquiry and, if necessary, again during the inquiry will reduce the likelihood of answers that stray into revelation of the jury’s thought process. The jurors can be instructed to respond about any information that was not mentioned during the trial (appropriate), but not to describe how the jurors used that information or the effect of that information on the thinking of any one or more jurors (inappropriate). Once any juror has established that extraneous information was mentioned, by whom, and whether anyone said anything else about the ex*392traneous information (not what they thought about it or did with it), the inquiry of that juror is complete. As soon as the judge determines that the defendant has satisfied his burden of establishing the existence of an extraneous influence, the questioning of all jurors should cease.
Nevertheless, some inappropriate information may be learned from the postverdict inquiry, as occurred here. If so, that information cannot be ignored. See Commonwealth v. Cuffie, supra at 638 (“The judge, having ventured into the forbidden area of the deliberative process . . . cannot ignore the fruits of his excursion”). If, as happened here, a judge learns that a juror has, in fact, been influenced by extraneous information, there must be a new trial. See id.6
3. Conclusion. The Commonwealth does not dispute that the judge in this case applied the correct legal standards, see Commonwealth v. Fidler, supra at 201. The Commonwealth points us to no Massachusetts case, and we have not located one, where an appellate court has reversed a trial judge’s finding of extraneous influence with resulting prejudice to the defendant after the judge applied the proper standards. While the facts of this case are close, it is in just such difficult cases that deference to the trial judge in assessing the credibility of the evidence before him is most appropriate.
The Superior Court’s order that the defendant be granted a new trial is affirmed.
So ordered.
The defendant was acquitted of drugging another with the intent of having unlawful sexual intercourse. G. L. c. 272, § 3.
The defendant filed a motion for a postverdict inquiry. It is unclear from the record whether he ever filed a motion for a new trial. In any event, he requested a new trial in a memorandum following the postverdict inquiry.
The barrette was described as a child’s “pony-tail holder,” consisting of “two little balls” with an “elastic.”
Lampron has since been located and the prosecution of his case is ongoing. See Jansen, petitioner, ante 112 (2005); Commonwealth v. Lampron, 441 Mass. 265 (2004).
As the judge here noted, our past cases suggest this standard. See, e.g., Commonwealth v. Guisti, 434 Mass. 245, 253 (2001); Commonwealth v. Fidler, 377 Mass. 192, 201 (1979). Immediately after stating that the defendant bears the burden of proving that the jury were exposed to extraneous matter, these cases state that the Commonwealth bears the burden of proving prejudice “beyond a reasonable doubt.” It is reasonable to infer that had we intended *387that the defendant’s burden was to prove the exposure to extraneous matter “beyond a reasonable doubt,” we would have stated so explicitly.
Here, even without considering the improper responses of juror A, the judge concluded that the Commonwealth had not met its burden of proving that the defendant was not prejudiced by the extraneous matter.