delivered the opinion of the Court.

This case tests the limits of a judge’s authority to question witnesses at trial. Trial courts have an obligation to insure that trials proceed fairly, and judges may question witnesses to clarify testimony, expedite a case, and otherwise protect the proceedings. But in exercising their discretionary power, judges must take care not to influence the jury by signaling doubt about a witness’s credibility. To do otherwise might place the court’s impartiality in question and affect the trial’s outcome.

In this case, the trial judge’s questioning of a criminal defendant ran counter to the above principles. Accordingly, we reverse defendant’s conviction and sentence and remand for a new trial.

I.

A.

Defendant Michael Taffaro and his sister Susan Taffaro were enmeshed in a dispute about their parents’ estate. While the[*446] particulars of the dispute are not relevant, defendant’s behavior resulted in the entry of a restraining order against him in January 2004. The order barred him from “any direct physical contact” with his sister (or a third sibling) and also directed defendant “not to communicate with [his siblings], either personally, or by telephone, in writing, or in any other manner directly or indirectly.” Because of pending litigation about their parents’ estate, the order allowed for only one exception: defendant’s “professional representatives” could contact his siblings’ “professional representatives,” if necessary.

On March 29, 2004, around midnight or 1:00 a.m., Ms. Taffaro received a phone call at her home. A strange voice identified himself only as Wesley and asked to speak to the “lady of the house.” He also asked whether she lived in a particular town in northern New Jersey where she, in fact, resided. Ms. Taffaro learned from the caller some details about how he came to contact her, and reported to the police the next morning that she was listed on an ad on the Internet. The police confirmed that a salacious posting had been placed on Craigslist inviting people to call the “lady of the house,” at Ms. Taffaro’s unlisted telephone number, for sexual favors.

Craigslist has the features of an old-fashioned bulletin board that is accessible through the Internet. People can post a variety of advertisements on Craigslist, including personal ads, using a computer connected to the Internet. The police investigation in this case tracked the posting to a computer linked to the defendant. As a result, on July 27, 2004, a Bergen County grand jury charged defendant with one fourth-degree count of contempt, contrary to N.J.S.A. 2C:29-9(a), for disobeying the restraining order by posting an ad on Craigslist.

At a two-day trial that began on February 28, 2005, defendant did not contest that the Craigslist ad was posted from his home computer. Rather, during a direct examination that lasted twenty-six minutes, defendant claimed that two acquaintances, Daniel Ng and Redner Pórtela, typed the illicit message while visiting him.[*447] Defendant testified that he had invited the two to his home to fix a virus on his computer and had served them dinner. Defendant claimed it was Ng’s idea to write the ad, and that Ng and Pórtela had access to his sister’s phone number and address because the information was beside the computer. When defendant heard the two laughing, he entered the computer room where they were working, and they read him the draft text of an ad. Defendant testified that he ordered Ng and Pórtela to remove the message immediately and that he believed them when they said they had done so. Defendant claimed not to have known that the ad was posted until his arrest on April 6, 2004.

Ng and Pórtela both testified to the contrary. They each explained that while they were at defendant’s apartment, he complained about his sister stealing his inheritance. Defendant told them he wanted to get back at her, and Ng suggested that defendant post something on the Internet. Ng showed defendant the site for Craigslist and told him how to use it. Afterward, both Ng and Pórtela testified that they were in the kitchen while defendant spent twenty minutes alone in the computer room. They denied typing or posting the offensive ad in question.

In light of the conflicting testimony, the case devolved into a battle over credibility. Not surprisingly, the prosecutor challenged defendant’s version of events during a brief, nine-minute cross-examination. She attempted to establish that defendant and his sister had a bad relationship, which provided a motive for the posting, that the restraining order prohibited indirect contact via e-mail, and that posting an ad on Craigslist would violate the order. As on direct examination, defendant denied composing the ad and affirmed that, as far as he knew, Ng and Pórtela deleted it at his request. In response, the prosecutor repeatedly pressed defendant as to whether he checked Craigslist to see if the message had been posted. Defendant countered that he would have done so had he known the message was posted, that he had not touched the computer, and that he did not even know which website to check.

[*448] During cross-examination, the prosecutor referred to the restraining order that permitted defendant to relay information to his sister through counsel. Specifically, the prosecutor asked whether defendant had informed his lawyer about the Craigslist before or after his order to warn his sister. Once again, defendant testified that he did not know the message had been posted until his arrest.

At the end of cross-examination, the trial judge proceeded to question the defendant for less than five minutes. The judge posed more than thirty questions overall, to which defense counsel did not object. The exchange established that defendant, Ng, and Pórtela did not have a particularly close relationship, and that it was important to the defendant that the message not be posted. In light of those facts, the trial judge asked, in various ways, why defendant trusted that Ng and Pórtela had deleted the message. Like the prosecutor, the judge asked defendant whether he checked to make sure the message was not sent. Defendant’s responses were consistent with his answers on cross-examination. All together, the period of time for the trial court’s questions and defendant’s responses amounted to half the length of the brief cross-examination.

The trial court also posed a few questions of another witness. After direct- and cross-examination, the court briefly asked Ng about his departure from defendant’s apartment and whether he had seen the posting on Craigslist before leaving.

In its instructions to the jury, the trial court explained that “[t]he fact that I asked a question of a witness ... must not influence you.... The fact that I asked a question does not indicate that I hold any opinion one way or the other as to the testimony given by that witness.”

The jury found defendant guilty of contempt. On April 29, 2005, the trial court sentenced defendant to one year of probation and placed him on notice that a breach of the restraining order would violate his probation. Defendant appealed.

[*449] B.

On appeal, defendant argued, among other things, that the nature and extent of the trial court’s questions were improper. By expressing doubt about defendant’s testimony, defendant claimed the trial court abandoned its neutrality and prejudiced defendant before the jury. Defendant also challenged the prosecutor’s cross-examination about defendant’s comments to his attorney.

In an unpublished per curiam opinion, the Appellate Division affirmed defendant’s conviction. The panel found no plain error in the trial court’s questions. While noting that the “questioning was somewhat extended,” the panel attributed that, at least in part, to defendant’s non-responsive answers. The panel found the court’s questions focused, neutral in tone, and non-demeaning. Although the exchange “diminish[ed] the weight of defendant’s testimony,” the panel concluded that it did not “render it adversarial, but merely probing.” In the context of the entire trial, the panel found that the judge had not failed to act impartially.

The panel also found permissible the prosecutor’s questions about defendant’s pre-arrest silence. The panel reasoned that the questioning focused on a period “considerably before” defendant’s arrest and “did not involve any interaction between defendant and the police.” Thus, the panel held that no violation of State v. Muhammad, 182 N.J. 551, 868 A.2d 302 (2005), occurred.

The State conceded it was error for the prosecutor to ask about defendant’s post-arrest silence. The Appellate Division found that “brief attack” on defendant’s credibility harmless.

This Court granted defendant’s petition for certification limited to two issues: (1) whether the trial court’s questioning of defendant was improper; and (2) whether the prosecutor violated defendant’s right to remain silent by asking about statements defendant did not make to his attorney. 191 N.J. 318, 923 A.2d 232 (2007).

[*450] II.

Defendant argues that the trial court’s questioning crossed the line that separates permissible judicial intervention in a trial from improper advocacy. Defendant contends that the trial court’s persistence telegraphed its personal view of defendant’s credibility, namely, that the judge thought defendant was untruthful. Particularly in a case that turned on defendant’s credibility, defendant submits that the court’s questions deprived him of a fair trial. Defendant also contends that the prosecutor violated his right to remain silent by cross-examining him about his failure to speak with his before and after his arrest-about the ad posted on Craigslist.

The State contends the Appellate Division correctly ruled that the trial court’s neutral questions did not overstep the bounds of judicial propriety. Moreover, the State submits that the trial court properly instructed jurors not to be swayed by the court’s questions. The State also argues that the prosecutor’s questions about defendant’s pre-arrest silence did not run afoul of Muhammad, supra, and that a concededly erroneous question about defendant’s post-arrest silence was harmless.

III.

A.

Trial courts can and should intervene at trial in certain circumstances. Above all else, judges must insure a fair trial. They should not sit silently when plainly improper tactics threaten a party’s basic rights. State v. Guido, 40 N.J. 191, 207, 191 A.2d 45 (1963). Judges may also intervene to expedite a trial and prevent delay or waste of time. Ibid.

N.J.R.E. 614 specifically authorizes judges to call or question witnesses “in accordance with law and subject to the right of a party to make timely objection.” Under our case law, it is entirely proper for judges to ask witnesses questions to clarify[*451] their testimony. Guido, supra, 40 N.J. at 208, 191 A.2d 45; Ridgewood v. Sreel Inv. Corp., 28 N.J. 121, 132, 145 A.2d 306 (1958). Similarly, when a witness is in severe distress, it may be appropriate for a judge to pose questions to help elicit facts. State v. Riley, 28 N.J. 188, 200-04, 145 A.2d 601 (1958).

During jury trials, though, trial courts should use great restraint in questioning witnesses in order not to influence the jury. Ridgewood, supra, 28 N.J. at 132, 145 A.2d 306. As Chief Justice Weintraub observed nearly a half century ago, trial judges are “imposing f!gure[s]”; to jurors, they symbolize “experience, wisdom, and impartiality.” Guido, supra, 40 N.J. at 208, 191 A.2d 45. Therefore, if a judge’s questions “suggest disbelief, the impact upon the jurors may be critical.” Ibid. This is especially true when the outcome of a case rests “primarily and necessarily” on the jury believing or rejecting a defendant’s version of events. State v. Ray, 43 N.J. 19, 25-30, 202 A.2d 425 (1964) (finding reversible error in trial court’s questioning of defendant, which signaled “doubts as to her credibility”).

In addition, while it is proper for judges to attempt to clarify testimony, they should not press defendants when the meaning of their responses is “perfectly plain.” Guido, supra, 40 N.J. at 208-09 n. 2, 191 A.2d 45. Questions of that sort may express incredulity and prejudice a defendant.

These concerns are less acute in the context of bench trials, where judges serve as fact finders and have more latitude in questioning witnesses. (State v. Medina, on which the Appellate Division relied, involved a bench trial. 349 N.J.Super. 108, 793 A.2d 68 (App.Div.), certif. denied, 174 N.J. 193, 803 A.2d 1165 (2002).) The critical concern, of course, is that a court not suggest to jurors through its questioning that it is taking one party’s side. To do so is to “cross the fine line that separates advocacy from impartiality.” Ridgewood, supra, 28 N.J. at 132, 145 A.2d 306.

[*452] B.

The outcome of this trial necessarily turned on the jury’s view of two opposing stories. The jury had to choose between defendant’s Ng and Pórtela posted the ad on Craig-slist against defendant’s Ng and Portela’s version that they neither prepared nor posted the ad. Thus, defendant’s conviction rested in large measure on whether the jury found his testimony credible.

As recounted earlier, defendant denied writing the ad, explained that he directed Ng and Pórtela to delete the draft message, and claimed that he had no idea the ad was posted until his arrest on April 6, 2004. The prosecution’s cross-examination was directed at undermining defendant’s credibility. In particular, the prosecutor asked a number of different ways whether defendant, who appreciated the seriousness of the situation, had bothered to cheek if the message was actually posted. Defendant provided consistent, plain answers emphasizing that he did not know the ad was on the Internet and would not know where to check in any event. While defendant’s answers may have been hard to believe, that issue was for the jury alone to decide.

The trial court then posed a series of questions that neither redressed the tactics of the parties nor helped expedite the trial. Even more important, they did not help clarify defendant’s understandable testimony; instead, they underscored the weaknesses in his defense.

At the outset, the trial court elicited, question by question, that Ng and Pórtela were not good friends of defendant’s; that he did not socialize often with them; that they were just acquaintances; that he would not lend them money; and that he had not seen them for months before they came to his apartment. Defendant next conceded, as he had earlier, that it was important for him that the message be deleted. The court’s questions then focused on why defendant had trusted Ng and Pórtela:

[*453] THE COURT: So then why did you trust, if they weren’t that close to you, and if it was important to you to know that it was off the screen, why would you trust two guys that you weren’t that close to to have gotten rid of it without checking?

DEFENDANT: Weil, because they said they took it off. So what they did is they went on to other sites. So on the screen, I see other things up there.

THE COURT: Well, did you actually go back to check to see if it was off?

DEFENDANT: No, because I’m guilty of would do something like that?

THE COURT: My question to you is, if you didn’t know them that well, then why would you trust that they would have done this without checking to make sure?

DEFENDANT: I wouldn’t have been able to even check. Because I don’t know— they said they took it off, I’m looking on the computer screen, it’s not there.

THE COURT: Did you call Ng the next day and ask him for the site?

DEFENDANT: No, I can’t call him. I have to wait for him to call me_

THE COURT: So when you woke up the next morning, did you think the thing was off the computer?

DEFENDANT: Absolutely, no reason for me not to. He knows the whole situation with my sister. I tell everybody.

THE COURT: But you saw what was on the screen, and you saw what he was intending to put on the screen, correct?

DEFENDANT: Right. But if you don’t send it, can type anything that’s—

THE COURT: How did you know he didn’t send it?

DEFENDANT: Well, I know now.

THE COURT: Well, how do you know he didn’t send it then?

DEFENDANT: Because he told me, he said we took it off.

THE COURT: Well, why would you trust him if he told you that if he’s a guy that you only saw once in awhile and you didn’t even have his phone number?

DEFENDANT: I don’t know.

As the videotape of the trial reveals, the trial court’s tone was neutral throughout the questioning. And the nature of defendant’s responses did, indeed, prolong the exchange. Regardless, the questions had the effect of suggesting to the jury that the court doubted defendant’s account in a case that rested heavily on defendant’s credibility. The questions also covered, in part, terrain that had already been crossed. Rather than clarify points in a witness’s testimony, the court’s questions had the capacity to signal disbelief. The overall length of the to half the time of the prosecutor’s brief cross-examination—[*454] compounded the error. But it is the impact of the court’s questions, and not the number of minutes they lasted, which matters most.

By way of contrast, the trial court’s brief questioning of Ng was not at all improper. The court clarified certain facts without suggesting any view.

C.

Because defense counsel did not object to the questions asked of defendant at trial, we review them under the plain error standard. R. 2:10-2. Applying that standard, an error is reversible if it was “clearly capable of producing an unjust result.” Ibid. In other words, was the possibility of injustice “sufficient to raise a reasonable doubt as to whether the error led the jury to a result it otherwise might not have reached”? State v. Macon, 57 N.J. 825, 386, 273 A.2d 1 (1971).

In light of the trial judge’s esteemed position in the courtroom and the central role that defendant’s credibility played in this trial, suggesting disbelief of defendant’s testimony could well have had a critical impact on the verdict. See Ray, supra, 43 N.J. at 26-30, 202 A.2d 425; Guido, supra, 40 N.J. at 207-08, 191 A.2d 45. As a result, the trial court’s questions warrant reversal. See R. 2:10-2. We are not persuaded that the jury instruction was sufficient to cure the harm. See Guido, supra, 40 N.J. at 208, 191 A.2d 45 (trial court’s questions were prejudicial despite judge’s “repeated assurances to the jury that he was acting in the interest of justice with no purpose of aiding or hurting the prosecution or the defense”).

To be clear, we find no improper intent on the part of the trial judge. Nonetheless, his questioning had the effect of depriving defendant of a fair trial, which requires reversal.

IV.

Defendant also challenges the prosecutor’s questions about his pre- and post-arrest silence. We discuss each area, in turn.

[*455] Under federal law, a defendant’s pre-arrest silence before receiving Miranda ^[1] warnings can be used to impeach his or her credibility. Jenkins v. Anderson, 447 U.S. 231, 238-40, 100 S.Ct. 2124, 2129-30, 65 L.Ed.2d 86, 95-96 (1980). Neither the right against self-incrimination guaranteed by the Fifth Amendment, nor the due process protections of the Fourteenth Amendment, is violated under those circumstances. Id. at 240-41, 100 S.Ct. at 2130, 65 L.Ed. 2d at 96.

The Supreme Court in Jenkins invited states to formulate their own rules of evidence on the subject. Ibid. Since then, this Court has spoken to the issue on a number of occasions. Under New Jersey law, a defendant’s pre-arrest silence can be used for impeachment purposes (1) “if that silence ‘ significantly’ preceded his arrest and did not arise in a custodial or interrogation setting,” and (2) if a jury could infer that a reasonable person in the defendant’s position would have come forward and spoken. Muhammad, supra, 182 N.J. at 571-72, 868 A.2d 302 (quoting State v. Brown, 118 N.J. 595, 610, 573 A.2d 886 (1990)); see also State v. Brown, 190 N.J. 144, 158, 919 A.2d 107 (2007) (“We continue to adhere to the view that when a defendant testifies, ‘pre-arrest silence may be admitted for impeachment purposes provided no governmental compulsion is involved.’ ”) (quoting Brown, supra, 118 N.J. at 613, 573 A.2d 886).

In this case, defendant was forbidden from contacting his sister under the terms of the restraining order and could only communicate through a “professional representative.” The prosecutor asked defendant whether he contacted his lawyer, to get word to his sister, on the night the message was posted. The questioning focused on defendant’s actions a week before his arrest and did not involve police interrogation or any governmental compulsion. In addition, a jury could infer that a reasonable person in defendant’s situation would have acted differently and tried to warn the victim that a salacious ad had been posted about her on the[*456] Internet. See Muhammad, supra, 182 N.J. at 571-72, 868 A.2d 302. Thus, although prosecutors should exercise great care when asking questions of this type, we do not find a violation under the facts presented. ^[2]

In responding to the prosecutor’s questions, defendant emphasized that he did not know of the posting until after his arrest. The State properly concedes it was error for the prosecutor to ask defendant one additional question about whether he contacted his attorney after being arrested. It is well-settled under federal and state law that a prosecutor may not use a defendant’s post-arrest silence against him. Doyle v. Ohio, 426 U.S. 610, 619, 96 S.Ct 2240, 2245, 49 L.Ed.2d 91, 98 (1976); Muhammad, supra, 182 N.J. at 568, 868 A.2d 302. At defendant’s retrial, that inappropriate inquiry should be avoided.

Y.

For the reasons set forth above, we reverse the judgment of the Appellate Division, vacate defendant’s conviction and sentence, and remand for a new trial.