dissenting.
In an atmosphere of saturating, vicious publicity, a Burlington County jury convicted defendant Ambrose Harris in Mercer County for the carjacking, kidnapping, rape, and murder of Kristin Huggins. The same jury sentenced him to die. At trial defendant actively contested his guilt, particularly his status as trigger-man and, hence, his eligibility for the death sentence. Defendant directly appeals as of right his convictions.
The primary issue that defendant raises in this appeal relates to the massive, inflammatory pretrial and midtrial publicity about the ease and its impact on his trial, his conviction, and his death sentence. Defendant also raises substantial claims concerning the adequacy of the jury voir dire as related to jury selection and jury taint, the failure to bifurcate the jury for the guilt and penalty phases of the trial, the trial court’s failure in the penalty-phase of the trial to exclude or neutralize inadmissible evidence introduced at the guilt phase, and the failure to inform the jury that in the event defendant were not sentenced to die, he would receive lengthy consecutive sentences for the robbery, kidnapping, and aggravated sexual assault convictions.
I conclude that defendant’s principal claims of error are valid and that their prejudicial impacts thwarted a fair trial. The Court *212should reverse defendant’s convictions and vacate his death sentence. I, therefore, dissent.
I
A sea of horrendous, sensationalistic, and unremittant publicity engulfed this prosecution. The Trentonian, a daily tabloid newspaper, was the primary, almost exclusive, source of this unabated torrent.
Defendant initially requested that the proceedings be closed to the media, that venue be changed, and that, if the court declined to change venue, a jury from Camden County be impaneled. In ruling on those applications, the court acknowledged that the Trentonian posed a threat to a fair trial. The court, however, rejected defendant’s requests. It nevertheless agreed to impanel a foreign jury from Hunterdon County.
On an interlocutory appeal from that ruling, the Appellate Division held that the impanelment of a foreign jury was appropriate but that the trial court, by failing to consider racial demographics and refusing to consider impaneling a Camden County jury, had abused its discretion. 282 N.J.Super. 409, 660 A.2d 539. On remand, the trial court, again rejecting defendant’s request for a change of venue or for the impanelment of Camden County jurors, selected, for convenience, a jury from Burlington County although the circulation of the Trentonian and Trenton Times was significantly lower in Camden County than in Burlington County and the racial demographics between Mercer County and Camden County were similar. The Appellate Division denied leave to. appeal that determination.
A.
Publicity about the case started up as soon as Kristin Huggins disappeared in December 1992. The Trentonian and Trenton *213Times devoted extensive coverage to this case.1
The majority of the articles in the Trentonian and the Trenton Times described the crime, the investigation, the victim, her family, the court proceedings, defendant, and defendant’s criminal record. However, the Trentonian, and to a significantly lesser degree the Trenton Times, focused on defendant with shocking sensationalism that went vastly beyond factual reporting. As soon as defendant was implicated in the crime, the Trentonian, through its headlines, editorials, and letters to the editor, viciously attacked defendant. Its articles on Kristin Huggins, constantly referring to her as “artist” and stressing repeatedly that she had begged for her life, dwelled on her funeral and her family’s grief. They were calculated to arouse enormous sympathy.2
Well before trial, the Trentonian ran cover headlines, in huge print, and article titles, often accompanied by large photos of defendant, such as “He Knows What He Did,” “Suspect a Loudmouthed Punk,” “A Daily Dose of Terror,”3 “Profile of a MONSTER: The man who killed Kristin Huggins committed his first *214rape as a teenager,” “From boy to beast,” “Huggins slayer terrorizes prison,” “He’s Satan in disguise,” “ ‘Satan’ Makes Prison Bloody, Fiery Hellhole,” “Expert says rape indictment will prompt jury to ... KILL HIM!,” “ Wild Animal’ Guaranteed 30 Behind Bars,” “Nevermind [sic] Konko, just call him maggot,”4 and “Stop calling Ambrose names, lawyers whine.”5 The Trento-nian frequently called defendant “Squirt Boy,”6 but it also employed other epithets, such as “useless savage,” “well known coward,” “annoying armpit sweat,” “executioner’s bait,” and “fruit of a faulty contraceptive.”
The Trentonian at one point quoted someone as stating: “No trial. He shouldn’t live. Just get on with it and fry him. No trial.” A Trentonian reporter, responding to defendant’s attempt to change his name to Konko Lumumba, stated: “[F]rankly, we couldn’t care less what Ambrose/Konko calls himself. Just spell it right on the toe tag.”
Direct and vicious attacks on defendant also appeared in the Trentonian’s page known as “BackTalk,” in which the tabloid publishes readers’ spoken opinions. The opinions selected for publication and the editorial responses grossly vilified defendant. For example:
Opinion: “Why don’t we have a public hanging on Trenton High’s Football Field? We can invite all the students. Heck, we can invite everybody from all around to see what happens to people who don’t care about other people. Why not give it a shot? I’ll pay for the rope!”
Response: “Nice try, but I was the first to offer to pay for the rope. You can buy the lumber for the gallows, though.”
*215Opinion: “I liked your response to the caller who said it costs more to execute someone than to keep them in prison. The comment about rope being cheap was very good.”
Opinion: “Let’s bring back the death penalty and let’s set up a lottery to get rid of monsters like Ambrose Harris, ok? Let me be the first to pull the switch on this guy-”
Response: “It turns out the hot squat [i.e., electric chair] isn’t so pricey either, provided you get the industrial rate from the electric company. Two or three bucks tops for all those amps and volts. What a deal.”
Opinion: "When ‘Creeping Justice’ called to complain about how long it’s taking to get around to trying the worthless piece of murdering scum BackTalk now refers to as ‘Squirt Boy,’ you said, Wes, but take solace in knowing that eternity is long and hell’s fires endless.’ Maybe so, Ed, but in the meantime we’re still paying for Squirt Boy’s room and board. These thugs enjoy prison too much and too long.” Response: “I was only trying to look on the bright side. Squirt Boy will be screaming soon enough, and he’ll be screaming forever and ever.”
The unsigned editorial page continued this egregious campaign:
Harris has to have a trial and he has to be provided the best representation taxpayers’ money can buy. That’s what our justice system requires. Then there will be the usual appeals and further appeals.
But someday, years from now, if there is justice, the last appeal will have been rejected and the last stay of sentence vacated.
On that day, Ambrose Harris — cold-blooded murderer — will be strapped to a prison gurney. A needle will be inserted into his arm and a lethal mixture of drugs will be injected into his veins.
Minutes later, one of the biggest pieces of human trash ever to blight Trenton’s streets will be gone, and the world will be a far better place for his passing.
The Trentonian ... at various times has referred to Harris as “Artist Slayer,” “Maggot,” and “Squirt Boy,” the latter in anticipation of the happy day when the jury does what’s right and sends him to the execution chamber to get the lethal injection he so richly deserves.
Well, Mr. Call [defendant’s attorney, who in his attempts to change venue, criticized the Trentonian’s coverage] can whine and moan all he wants about how his pet maggot is too much in the spotlight, but the truth is he put himself there by capping his criminal career with the rape and murder of Kristin Huggins. It may offend Call’s sensibilities to hear his scumbag client referred to as “Artist Slayer” and “Squirt Boy,” but an artist slayer he is and a squirt boy he will be if the jury does its duty. Like it or not, that’s the way things are, Law Boy.
*216The Trentonian consistently portrayed defendant as negatively as possible. For example, the Trentonian repeatedly referred to defendant as “artist killer” and “career criminal Ambrose Harris.” It titled one article “Life of a career criminal” and another “Life and crimes of Ambrose Harris.”
In addition, by using large sensationalistic quotes that implied the statements were those of individuals — such as “He did it,” “Grieving parents: All murderers must die,” and “Ambrose a terror” — on its covers and in the titles of articles, the Trentonian injected inflammatory editorial comment and opinion in its ostensible factual reporting.
The Trentonian laced its factual reporting with references to defendant as a sociopath. It “reported” that “The Killer hates to hunt alone,” that “Ambrose Harris sees ‘murder as a joke,’ ” and that “[w]hen scenes of violence flicker across the television sereen[,] ... the alleged killer and rapist smiles.” It also reported that fellow inmates and prison guards hated defendant.
The Trentonian ran a number of reports on opinions regarding defendant. One such article, entitled “People agree Harris ‘shouldn’t live,’ ” stated:
Ambrose Harris should die: That was the simple message yesterday from the young barmaid, the grizzled war vet, the guys from the ice house and the stylish secretary.
All across Greater Trenton, people called for the death penalty — even torture— for the suspected killer of Bucks County artist Kristin Huggins, whom Harris is accused of raping and killing on Dec. 17,1992.
“There’s no excuse for a human like that. He’s a waste of space,” said Annie Summers, a barmaid in her early 20s. Her call for speedier justice and longer sentences won support at Jule’s Tavern in Trenton’s Chambersburg.
“He should have been Med a long time [ago]. He’s been a criminal [his.entire] life,” said John Shaffer, 42, a Vietnam veteran who works at an ice house in Trenton. One of Shaffer’s colleagues volunteered to put Harris on ice, so to speak.
“Let me give him the squirt (lethal injection),” said the man, who identified himself only as Paul L. “For all the suffering he put those girls through ... they should make him suffer for a while.”
Several people offered various modes of execution. A 62-year-old secretary from Trenton who offered one such idea said it should be done “at high noon on Broad and Market.”
*217The American Civil Liberties Union wouldn’t be happy to hear all the calls yesterday for simply shooting Harris, who has yet to stand trial for the Huggins murder.
“No trial,” said an 81-year-old Burg man who fought in the Battle of the Bulge.
“He shouldn’t live. He’s no good. Just get on with it and fry him. No trial,” said the WWII vet and retired city streets worker who wouldn’t give his name.
Apart from the charges in the present case, defendant had an extensive record that included violent offenses. The Trentonian and the Trenton Times,, often referring to him as a “career criminal,” a “one man crime spree,” a “troll” who had engaged in a “rape spree,” and the like, repeatedly publicized defendant’s record in headlines and article titles. Most egregiously, on the first day of the penalty phase, the Trentonian published a glaring, front-page headline that revealed: “Ambrose eyed in ’67 slay: Remains prime suspect in unsolved murder.”
Some of the midtrial publicity was racially inflammatory. It dwelled on the fact that an urban African-American man raped and murdered a suburban white woman. It recounted testimony that defendant referred to Huggins as a “white bitch” and had stated that he would spare a black robbery victim but kill a white one. The Trentonian reported an alleged story about defendant’s jailhouse campaign of terror against white inmates. It stressed that he had threatened another rape victim by allegedly telling her that she should not resist because he had “just killed a white girl.” The Trentonian and other papers also stressed defendant’s purported comment about having “knocked off some white girl.”
Much of the publicity was clearly intended to influence the jury. For example, after the guilt-phase jury commenced its deliberations, it initially deadlocked over the charge of aggravated sexual assault. The Trentonian ran a cover stating “One juror stalls verdict” and reported about an alleged holdout juror who “could save Harris’ life.” Another headline remarked: “Dawdling Harris jury draws public’s fire.” After the trial court instructed the deadlocked jury to continue its deliberations, the jury convicted defendant of all the charges. The Trentonian ran the following front-page headline: “GUILTY! So why’s this killer smiling? *218Because he’s seen juror No. 7 crying, and he thinks she’ll never go for the death penalty.” That reporting occurred while the unse-questered jury, which could leave the courthouse during lunch and which was not individually voir dired about exposure to publicity, continued to deliberate over a period of days. See discussion, infra at 225-30, 716 A.2d at 514-16. In a particularly vicious editorial, entitled “Death for Harris,” the Trentonian published an unsigned open letter to the jury that urged it to sentence defendant to death.
The publicity was intensely concentrated in Trenton. The Trentonian is present throughout and around Trenton, the site of the crime and the trial. A plethora of streeteorner vending machines and “hawkers” sold the Trentonian in Trenton, including the area surrounding the Mercer County Courthouse. Due to the ubiquity of the Trentonian and the immense font used for the front-page headlines, the Trentonian’s cover page is inescapable in this State’s capital.
B.
In order to guarantee a criminal defendant the right to a fair trial by an impartial jury, Irvin v. Dowd, 366 U.S. 717, 722, 81 S.Ct. 1639, 1642, 6 L. Ed.2d 751, 755 (1961), the trial court must take sufficient precautions to minimize adverse pretrial and mid-trial publicity that has the capacity to infect juror perceptions of the case. State v. Williams, 93 N.J. 39, 63, 459 A.2d 641 (1983) (Williams I). Those precautions are necessary if a “realistic likelihood of prejudice from pretrial publicity” exists. Id. at 67-68 n. 13, 459 A.2d 641; accord State v. Bey, 112 N.J. 45, 77, 548 A.2d 846 (1988) (Bey I). There can be no doubt that the pervasive pretrial and midtrial publicity that surrounded this prosecution created a realistic likelihood that defendant would not receive a fair trial. The precautions taken by the trial court to overcome that publicity were woefully inadequate to assure a fair trial.
In State v. Biegenwald, 106 N.J. 13, 524 A.2d 130 (1987) (Biegenwald II), the Court recognized the distinction “between *219eases in which the trial atmosphere is so corrupted by publicity that prejudice may be presumed and cases in which pretrial publicity, while extensive, is less intrusive, making the determinative issue the actual effect of the publicity on the impartiality of the jury panel.” Id. at 38, 524 A.2d 130 (citations omitted).
In determining whether pretrial and midtrial publicity lead to a presumption of prejudice, several factors including evidence of extreme community hostility, prominence of the victim and/or the defendant, the nature and extent of the coverage, the size of the community, the nature and gravity of the crime, and the temporal proximity of the publicity to the trial, are relevant. State v. Biegenwald, 126 N.J. 1, 23, 594 A.2d 172 (1991) (Biegenwald IV); State v. Koedatich, 112 N.J. 225, 272-73, 548 A.2d 939 (1988) (Koedatich I). If the court determines that a presumption of prejudice does exist, then it must take curative steps without regard to its belief about the actual effect of the publicity on jurors.
In this case, both the trial court and the Appellate Division agreed that a presumption of prejudice existed. 282 N.J.Super. at 413-15, 660 A.2d 539. Given the sheer volume, intensity, and hatefulness of the Trentonian’s purported journalism in this case, the Appellate Division correctly held that the trial court had not abused its discretion in concluding that a presumption of prejudice existed. According to the Appellate Division,
there is more than adequate support in the record for the trial court’s findings that the Trentonian, as part of a “vengeance seeking crusade” against defendant, has published a “stream of invective” that has been “constant,” “prolonged” and “sensationalized," that there is a “likelihood of its taint permeating the trial,” and that “the Trentonian will continue to foster vengeance.”
[Id. at 415, 660 A.2d 539.]
The majority concurs: “There can be no doubt that this case was accompanied by widespread, inherently prejudicial pretrial media coverage. Strong measures were ‘necessary to overcome the realistic likelihood of prejudice from pretrial publicity.’ Williams I, supra, 93 N.J. at 67 n. 13, 459 A.2d 641.” Ante at 145, 716 A.2d at 469.
*220The pretrial publicity in this case was in a different class from that experienced by the defendants in Koedatich I, supra, 112 N.J. at 272-73, 548 A.2d 939 (concluding presumption of prejudice did not arise because there was lack of evidence of community hostility against defendant, neither defendant nor- victim was prominent in the community, the victim was not public servant, defendant was not outsider, the articles did not assume defendant’s guilt, and there was two-year lapse between most intense publicity and trial) and Biegenwald II, supra, 106 N.J. at 35, 524 A.2d 130 (holding no presumption of prejudice arose because the publicity had been concentrated six months before trial). Many of the articles in this case unqualifiedly assumed and proclaimed defendant’s guilt and stridently called for his death. The articles literally flooded Mercer County before and during the trial. See Coleman v. Kemp, 778 F.2d 1487, 1538-40 (11th Cir.1985) (annulling through writ of habeas corpus, defendant’s conviction and death sentence for vicious killing of a family and basing its decision on the trial court’s failure to grant change of venue in light of presumption of prejudice from saturation-level publicity, particularly publicized assumption of guilt by law enforcement, revelation of inadmissible evidence, hostile atmosphere, and calls for death), cert. denied, 476 U.S. 1164, 106 S.Ct. 2289, 90 L. Ed.2d 730 (1986).
Because the extensive, vicious publicity prejudiced defendant, the trial court had a constitutional obligation to take adequate measures to ensure that defendant received a fair trial. In the face of the compelling circumstances, those measures required a change of venue. See Rideau v. Louisiana, 373 U.S. 723, 727, 83 S.Ct. 1417, 1419-20, 10 L. Ed.2d 663, 665 (1963) (stating in ease in which capital defendant’s videotaped confession was televised that “we do not hesitate to hold ... that due process of law in this case required a trial before a jury drawn from a community of people who had not seen and heard [defendant’s] televised ‘interview.’ ”); Biegenwald II, supra, 106 N.J. at 34, 524 A.2d 130; Coleman, supra, 778 F.2d at 1541 & n. 25 (concluding that although change of venue is the normal remedy to counteract presumption of *221prejudice, other remedies, such as thorough voir dire, can rebut presumption); Mayola v. Alabama, 623 F.2d 992, 1000-01 (5th Cir.1980) (same), cert. denied, 451 U.S. 913, 101 S.Ct. 1986, 68 L. Ed.2d 303 (1981). This Court has recognized the need for a venue change when prejudicial publicity infects a capital trial. “While we recognize that a change of venue may disrupt court administration, it is imperative that the defendant in a capital case receive an impartial jury. Trial courts should not be reluctant to grant motions for change of venue in capital cases. On the contrary, courts should grant such motions liberally.” Koedatich I, supra, 112 N.J. at 282, 548 A.2d 939.
Although the Court has not generally found that failure to change venue in capital cases constitutes reversible error, those cases did not involve either the extensive or vicious quality of newspaper coverage present in this case that gave rise to the presumption of prejudice. See State v. Marshall, 123 N.J. 1, 73-78, 586 A.2d 85 (1991) (Marshall I), cert. denied, 507 U.S. 929, 113 S.Ct. 1306, 122 L. Ed.2d 694 (1993); Koedatich I, supra, 112 N.J. at 282, 548 A.2d 939; Biegenwald II, supra, 106 N.J. at 36-37, 524 A.2d 130. If the refusal to change venue in Koedatich I was error, albeit not reversible error because there was no presumption of prejudice, then the failure to change venue in this case, which carries a presumption of prejudice, was reversible error.
I find no escape from the conclusion that a presumption of prejudice requires a change of venue and that, in the absence of a venue change, a reversal of defendant’s convictions and vacatur of his sentence must follow.
The majority, noting “the empanelment of foreign jurors was the first trial management technique that Williams I suggested to combat the effects of preexisting prejudicial pretrial publicity,” concludes otherwise. Ante at 146, 716 A.2d at 470. In that ease, the Court enumerated several options for a trial court to employ, including change of venue, impanelment of a foreign jury, and augmentation of the jury pool, in order to contain the effects of adverse publicity. 93 N.J. at 67, 459 A.2d 641. The Court also *222stressed the need for adequate voir dire to detect latent bias. Id. at 68, 459 A.2d 641.
The Court acknowledges merit in defendant’s argument “that because the goal is to ‘minimize the danger that prejudice [from extensive pretrial publicity] will infiltrate the adjudicatory process,’ Koedatich I, supra, 112 N.J. at 268, 548 A.2d 939 (quoting Williams I, supra, 93 N.J. at 63, 459 A.2d 641), the most effective method of minimizing the potential was to select a jury from a county which was outside of the circulation range of the Trenton newspapers.” Ante at 149, 716 A.2d at 471-72 (alterations in original). The Court concludes that “[t]he trial court took firm steps to ensure that none of those households that received the Trentonian (the newspaper containing the most inflammatory material) would be on this jury” and that “[i]t made little difference whether the jurors were from Burlington or Camden counties.” Id. at 150, 716 A.2d at 472. I strongly disagree with that conclusion.
The trial court attempted to cure the presumed prejudice arising from pretrial publicity merely by empaneling a jury from Burlington County and by inquiring superficially about exposure to pretrial publicity during jury selection. The court’s deficient response failed to overcome the effects of the overwhelming prejudicial publicity.
The trial court’s general approach to pretrial publicity during voir dire was to inquire into the subject only if a juror had not been excused for a more fundamental reason, such as the inability to be death-qualified. When the court did proceed to the pretrial-publicity issue, it simply asked, often in leading form, whether the venireperson was a reader of Trenton newspapers or whether he or she had other knowledge of the case. If the venireperson answered negatively, the questioning ended. The court did not probe further about what the venireperson may have heard or read from others who had read about the case. For example, the court queried: “And I gather you never read the Trenton papers?” That question was, however, neither penetrating nor *223extensive. Not surprisingly, the venireperson responded: “No, I never read them.” One potential juror who read the Trentonian “regularly” incredulously professed not to know anything about the prosecution. As was the case with all voir dire after the second day of jury selection, the attorneys were not permitted to ask him questions. The voir dire ended without any examination of the rather incredible string of assertions by the venireperson that, although he had read the Trentonian and Trenton Times on a regular basis, he never even had heard of Ambrose Harris or any other aspect of the case. However, when a venireperson answered the publicity-exposure question affirmatively, then the court asked additional questions. In another such situation, despite the venirepersoris clear knowledge of the case, her specific knowledge, gleaned from a Trentonian article, about defendant’s use of Huggins’s MAC card, and her admission that she had seen Trentonian headlines on the subject, the court barely explored what exactly she had seen and how that may have affected her.
When a venireperson communicated that she possessed some sort of knowledge about the ease, the court engaged in a superficial, and often leading, discussion about the nature of the venire-person’s exposure and failed to explore whether the venireperson had been exposed to the nastier and more inflammatory side of the Trentonian’s coverage.
Three of the seated jurors stated that they never had heard of the case before coming to court; the trial court did not attempt to delve much further with those jurors. Four jurors had a small degree of knowledge of the case, and the court generally barely evaluated that knowledge. Five seated jurors, however, demonstrated a potentially greater degree of knowledge or impression, to which the trial court responded with a slight degree of probing in the form of highly leading questions. One juror, for example, who, as indicated from the jurors’ questionnaire, had “some knowledge about the defendant Ambrose Harris, or the facts of the case” was asked by the court: “And I gather, I don’t want to put words in your mouth, but with the next question you indicate that *224you knew nothing about a 22 year old artist who was missing and later found in a shallow grave; is that correct?” He answered: “That is correct.”
Nor did the court pry into potential further knowledge about defendant’s criminal record, which even the most cursory read of an issue of the Trentonian dealing with the case would reveal, especially given its repeated references to defendant as “career criminal Ambrose Harris.” See ante at 216, 716 A.2d at 509. If a venireperson possessed some knowledge about the case, there was a reasonable chance that he or she knew, consciously or subconsciously, about other inadmissible aspects of the case.
The court’s elimination of attorney participation contributed to the inadequacy of the voir dire. Defense counsel clearly perceived the problem of exposure of Burlington County jurors to the Trentonian and expressed to the court the comment of a venire-person about the high level of knowledge in Burlington County about the case. The trial court’s attitude in that respect became particularly troublesome when it ended attorney-conducted voir dire and conducted its own limited questioning concerning pretrial publicity. The court took those steps in spite of defense counsel’s clear statement that he intended to focus his voir dire questioning on exposure to pretrial publicity. The court’s solution was to concentrate on the issue of exposure to pretrial publicity “from this point on.” Thus, in response to a venireperson’s expressed knowledge of the case and statement that many of the Burlington County jurors had knowledge of the case precisely because of Burlington County’s proximity to Mercer County and the employment of many Burlington County residents in Mercer County, the trial court simply concluded that this fact was basically irrelevant because there would also be publicity in Cape May or Sussex County.
Even though the Court concludes that a presumption of prejudice can be overcome without changing venue, the trial court in this case failed to take sufficient precautions to counteract pretrial publicity. See Coleman, supra, 778 F.2d at 1542 (concluding trial *225court’s leading questions during voir dire did not overcome presumption of prejudice). Because Burlington County borders Mercer County, many residents of Burlington County work in or around Mercer County. Moreover, the Trentonian and Trenton Times each have circulations of about 11,000 in Burlington County. Consequently, the impanelment of the foreign jury from Burlington County, in light of the inadequacy of voir dire, did not provide an adequate substitute for a change of venue and did not rebut the presumption of prejudice. The pretrial publicity requires reversal of defendant’s convictions and vacation of his death sentence.
C.
The trial court also failed to counteract the effects the substantial midtrial publicity had on the jury. Defense counsel understandably was concerned about the effects of the midtrial publicity on the jurors and consequently proposed several prophylactic measures to the court. He proposed rerouting the juror bus and escorting jurors into the courthouse. The trial court rejected those proposals, even though they were unopposed by the State. The trial court also rejected defense counsel’s request to require the jury to remain in the jury room during lunch, to sequester the jury during the penalty phase, and to voir dire the jurors individually during the penalty phase. Instead, the court only admonished the jurors each day not to read or watch anything dealing with the case; on three separate occasions, it also asked the entire jury in open court whether it had been exposed to publicity, and, on each occasion, the jurors shook their heads “no” in unison.
During the trial, the TrentoniarCs wrath was typically directed at defendant. Nevertheless, the jury did not escape the tabloid’s ire. After the jury was unable to reach a verdict on all counts after the first day of guilt-phase deliberations, the Trentonian’s cover headline read: “One juror stalls verdict.” Page three *226articles were entitled “Dawdling Harris jury draws public’s fire”7 and “Lone juror could save Harris’ life.” The jury was not spared the day after it convicted defendant of all counts; the Trentoni-an’ s front-page headline, appearing adjacent to a photograph depicting defendant grinning, announced: “GUILTY! So why’s this killer smiling? Because he’s seen juror No. 7 crying, and he thinks she’ll never go for the death penalty.” That week the Trentonian also reported that juror number five was a Quaker, and the newspaper feared that his religion’s tenets opposing capital punishment would preclude him from voting to sentence defendant to die.
The midtrial publicity was most volatile on the first day of the penalty phase. That day, the Trentonian ran a cover headline implicating defendant in a 1967 murder. The headline, in large type, proclaimed: “Ambrose eyed in ’67 slay: Remains prime suspect in unsolved murder.” On page three, the newspaper printed an article reporting that the son of a woman killed in 1967 believed that defendant had committed the murder. It was also in this issue that the Trentonian printed an editorial that urged the jury to sentence defendant to death.
In response to this inflammatory issue of the Trentonian, defense counsel urged the trial court to voir dire each juror individually. That morning the court, declining to conduct an individual voir dire, merely asked the jury en banc if anyone had been exposed to newspaper articles or headlines and, noting the risk that Trenton newspapers could be seen in the courthouse, admonished the jury to avoid newspaper coverage of the trial. That was only the third time the court conducted an en banc voir dire concerning exposure to midtrial publicity. The court also permitted the jury to leave the courthouse during the lunch *227recess. In addition, the court never again inquired whether any juror had been exposed to the midtrial publicity.
In Bey I, supra, 112 N.J. at 79, 548 A.2d 846, the trial generated substantial publicity, much of which contained inadmissible information about the defendant, including his prior convictions and his pending indictment for another murder. This Court held that the trial court had abused its discretion in relying only on an admonition without polling the jury about exposure. Id. at 81, 548 A.2d 846. The Court stated that individual voir dire of jurors outside of the presence of other jurors was “likely to be more effective in uncovering any exposure than is questioning the jury era banc, in open court.” Id. at 86-87 n. 26, 548 A.2d 846.8 The Court concluded by stating:
The procedure of questioning an impaneled jury when prejudicial publicity threatens the fairness and integrity of a defendant’s trial should not be invoked begrudgingly. While we do not mean to suggest that any publicity relating to the defendant or the proceedings will automatically require that the jury be polled, a court might properly choose to err on the side of caution when ruling on such motions. The procedure is prophylactic in nature, designed to uncover potential prejudice to extremely significant constitutional rights that might otherwise go wholly undetected, and to do so at a time when corrective measures remain possible, that is, before ordering a new trial has become the only option____ “[W]e must remember that reversals are but palliatives; the cure lies in those remedial measures that mil prevent prejudice at its inception.”
[Id. at 89-90, 548 A.2d 846 (citations omitted) (quoting Sheppard v. Maxwell, 384 U.S. 333, 363, 86 S.Ct. 1507, 1522, 16 L. Ed.2d 600, 620 (1966)) ].
The majority, noting simply that unlike the Bey jurors who were admonished, the “Harris’ jurors were questioned generally and that inquiry revealed that no exposure had occurred,” ante at *228153, 716 A.2d at 473, distinguishes Bey I, supra, and therefore concludes that “a denial of individual voir dire should not form the basis for reversing a conviction when there is no evidence of exposure,” ante at 153, 716 A.2d at 473-74.
The Court’s reasoning is flawed. The absence of direct evidence that the jury was exposed to the publicity, which was unavoidable in Trenton, can be attributed to the lack of individualized voir dire. Due to the trial court’s woefully inadequate effort to uncover jury exposure to publicity, the Court’s reliance on the absence of direct proof that the jury had been exposed to publicity is disingenuous and misplaced. Moreover, under Bey I’s two-part test, the trial court in this ease was enjoined to take vigorous prophylactic measures to avoid juror exposure to midtrial publicity. First, the publicity given to defendant’s criminal record and to the unsubstantiated allegations of his perpetration of a 1967 homicide, the calls for his conviction and execution, and the attempts to encourage jurors to convict and to sentence defendant as quickly as possible were calculated to influence the jurors and had the capacity to prejudice defendant. Second, given the sheer volume of coverage, the prominence of the Trentonian’s cover stories, sales of the Trentonian and Trenton Times in and around the courthouse, and the newspapers’ substantial circulations in Burlington County, there was a “realistic possibility that such information may have reached one or more of the jurors,” Bey I, supra, 112 N.J. at 86, 548 A.2d 846.
Sequestration during the penalty-phase deliberations probably would .also have been an effective means of shielding the jury during those charged deliberations. Rule 1:8 — 6(b) allows the trial court, at its discretion, to sequester a civil or criminal jury during deliberations. Due to the massive midtrial publicity, the court should have sequestered the jury during penalty-phase deliberations.
At the very least, the court should have barred the jurors from leaving the jury room during lunch and modified their bus route and entry into the courthouse to avoid exposure to the Trentoni-*229an’s graphic and inflammatory covers. Although the jurors presumably were conscientious in attempting to avoid exposure to midtrial publicity, we cannot be confident that none was exposed, given the utter lack of preventive measures taken by the trial court. It is difficult to imagine that none of the jurors ever were exposed to prejudicial midtrial publicity. The jurors began their day by being bussed to the courthouse, probably passing by the Trentonian’s huge headlines, and by entering the courthouse without being guided away from the tabloid. By the time they were seated each morning, therefore, a significant possibility existed that one or more of them had been exposed either consciously or subconsciously to some sort of publicity regarding either defendant or the jury itself. At lunch, the jurors were permitted to wander around the courthouse and the downtown area; this increased the chance of exposure. During the entire trial, including the penalty phase, they went home each night, where they had yet another opportunity for inadvertent exposure. Yet, when asked by the trial court on three occasions whether they had been exposed, they failed to indicate that they had. Of course, that failure is not surprising given the attendant embarrassment to such an open-court admission.
Moreover, the one precautionary measure that the trial court actually took — occasional voir dire of the jury regarding potential exposure — was insufficient to neutralize the prejudice created by potential exposure to midtrial publicity. Although this Court in Bey I, supra, refused to create a blanket rule of individualized voir dire of jurors in order to gauge exposure, it strongly intimated its approval of such a procedure, 112 N.J. at 86-87 n. 26, 548 A.2d 846, and cited favorably to the ABA standard that recommends individualized voir dire. Id. at 87-88 & n. 28, 548 A.2d 846.
To rely exclusively on admonitions and occasional en bane voir dire in open court to discover, to gauge, and to counteract potential juror exposure to the massive and inflammatory midtrial publicity in this case was a seriously inadequate response. The trial court’s consistent refusal to take serious steps to address the *230midtrial-publicity problem — indeed, it even refused to take measures that the prosecutor did not oppose — is almost impossible to understand or to explain, and the Court does not do so. Because the trial court failed to discharge its duty to take adequate precautionary measures to minimize the prejudice flowing from midtrial publicity, the Court, on this basis alone, should reverse defendant’s conviction. At the very least, the Court should vacate his death sentence.
D.
The majority clearly recognizes the extreme danger posed by the publicity surrounding this prosecution:
Because in eases involving the death penalty a trial court’s responsibility under both the federal and state constitutions is to “minimize the danger that prejudice will infiltrate the adjudicatory process,” State v. Williams, 93 N.J. 39, 63 [459 A.2d 641] (1983) (Williams I), we hold that when hereafter there is a reasonable likelihood that the trial of a capital case will be surrounded by presumptively prejudicial media publicity (as that phrase is understood in the law) the court should transfer the case to another county. Other devices, such as restraints against the publication of material concerning the trial or the sequestration of jurors, have proven either to be unavailable to counter the effects of continuing prejudicial publicity or to produce a contrary effect than desired. In some cases a court may conclude that an initial tide of inherently prejudicial publicity will have subsided at time of trial and will not require a change of venue if the jury selection process yields an impartial jury. E.g., State v. Koedatich 112 N.J. 225, 273-82 [548 A.2d 939] (1988), cert. denied, 488 U.S. 1017, 109 S.Ct 813, 102 L. Ed.2d 803 (1989) (Koedatich I). When, however, a court is satisfied that there is a reasonable likelihood of the continuing recurrence at a capital trial of presumptively prejudicial publicity that might infiltrate the trial, a change of venue is required.
[Ante at 133-34, 716 A.2d at 463.]
Understanding the necessity of a venue change to preserve a fair trial in cases, such as this one, that are swamped by prejudicial publicity, the Court prospectively rules:
In future capital cases a court should change the venue of a capital trial when there is a realistic likelihood that presumptively prejudicial publicity will continue during the conduct of a trial. Presumptively prejudicial publicity is recognized as a barrage of inflammatory reporting that may but need not include all of the following: evidence that would be inadmissible at the trial, editorial opinions on guilt or innocence, and media pronouncements on the death-worthiness of a defendant.
[Ante at 147-48, 716 A.2d at 471.]
*231That belated and anemic concession of the trial court’s error has a dull and hollow ring. It is too little and too late for defendant, who was sentenced to die by a jury that was inescapably exposed to highly prejudicial publicity.
II
The voir dire in this case was wholly insufficient and failed to lead to the selection of an adequately death-qualified jury and was grossly inadequate in failing to overcome the high risk of racial bias.
A.
Jury selection began on October 10, 1995. The trial court and parties agreed that the court-conducted voir dire would be followed by attorney follow-up voir dire. However, on October 16, the trial court terminated attorney-conducted voir dire. The court reasoned that New Jersey had characterized attorney-eon-ducted voir dire in death-penalty cases as a “vestige” and that “all of the evils that the 1969 rule change sought to eradicate,” including juror “shopping,” remained in capital cases. The court, noting the “exasperation [of jurors] at the incomplete or twisted questions thrown at them” by defense counsel, expressed regret at having excused the two jurors who had been “confused” by defense counsel’s hypothetieals. It limited counsel to inquiries “only for purposes of clarification.” Counsel could not ask “hypothetical questions which are not strictly based on the law.” As a result, the attorneys were no longer permitted to ask any questions. Further, defendant exhausted his allotment of peremptory challenges and was denied additional peremptories.
Although the determination of whether to allow attorney-conducted voir dire rests in the sound discretion of the trial court, State v. Pennington, 119 N.J. 547, 591, 575 A.2d 816 (1990); State v. Hunt, 115 N.J. 330, 347-48, 558 A.2d 1259 (1989); State v. Williams, 113 N.J. 393, 426-27 n. 10, 550 A.2d 1172 (1988) (Williams II); State v. Zola, 112 N.J. 384, 394-97, 548 A.2d 1022 *232(1988); Koedatich I, supra, 112 N.J. at 291-92, 548 A.2d 939; Biegenwald, II, supra, 106 N.J. at 28-30, 524 A.2d 130, in capital cases trial courts must be especially sensitive to suggested defense voir dire questions, Williams II, supra, 113 N.J. at 426-27 n. 10, 550 A.2d 1172; Biegenwald II, supra, 106 N.J. at 30, 524 A.2d 130. This Court has encouraged active attorney participation in voir dire. See Pennington, supra, 119 N.J. at 591, 575 A.2d 816.
The Court now validates the jury selection process that was followed in this case and concludes that the voir dire permitted the selection of a adequately death-qualified jury. Ante at 161-70, 716 A.2d at 477-82. I disagree.
The trial court excused three jurors out of seven for cause as a result of the attorney-conducted voir dire that was initially allowed. The trial court’s accusations — that defense counsel was attempting to sabotage the process and was aggressively going after perceived pro-death jurors — were misplaced. Defense counsel was entitled to attempt to reveal the disqualifying opinions; that is the purpose of voir dire. Moreover, defense counsel’s questions were not misleading. Unlike the Court’s questions, they consisted of open-ended inquiries.
When the trial court terminated attorney-conducted voir dire, it permitted the parties to submit questions for the court to ask prospective jurors. Throughout the voir dire process, defendant, in written form, proposed many questions relating to a wide variety of subject areas. The trial court, however, asked few of his proposed questions.
The majority, indicating that “the [trial] court provided a ‘thorough and searching inquiry' into the jurors’ attitudes and biases,” treats the ruling to eliminate and reduce attorney voir dire dismissively. Ante at 163, 716 A.2d at 478.
Before the termination of attorney-conducted voir dire, defense counsel spent considerable time with prospective jurors about their views on capital punishment. Once the trial court terminated attorney-conducted voir dire, it generally conducted, over *233defense counsel’s repeated and strenuous objection, death-qualification voir dire in a leading manner. In its voir dire, the court went through its routine of describing the death-penalty process and asking extended, leading questions to which the juror responded in the affirmative. Defense counsel objected to this superficial form of voir dire. Despite uncertainty and equivocation expressed by many of the prospective jurors, the court simply and uncritically qualified them.
In the voir dire of several prospective jurors who eventually were seated in the case, the jurors often initially responded that death was appropriate only for some murders. Upon further questioning about which types of murders were inappropriate for death, they responded that it was inappropriate for accidental deaths, self-defense, or some other form of nonmurder. The court then explained the definition of murder and again asked the juror what type of murder would be deathworthy. The juror would then respond that “vicious,” “cold-blooded,” “premeditated,” “higher level,” or some other type of intentional murder deserved the death penalty. Finally, the court would describe, in a leading manner, the mechanics of the death penalty in New Jersey and elicit affirmative responses from the jurors about their ability to follow the law.
In Biegenwald IV, supra, 126 N.J. at 36-43, 594 A.2d 172, this Court criticized the trial court for the leading nature of its questions, its failure to probe problematic responses, its frequent exclusion of defense counsel from the questioning process, and its failure to explain the definition of “murder” to obviously confused venirepersons. In Williams II, supra, 113 N.J. at 415-17, 550 A.2d 1172, this Court termed the trial court’s refusal to go into further depth about the effect of other crimes on jurors “serious error.” See also State v. Martini, 131 N.J. 176, 211-12, 619 A.2d 1208 (1993) (Martini I) (holding, if kidnapping is charged in addition to capital murder, trial court must inquire into effect of kidnapping charge on prospective jurors), cert. denied, 516 U.S. 875, 116 S.Ct. 203, 133 L. Ed.2d 137 (1995). In this case, the trial court committed the same error committed in Biegenwald IV and *234Williams II. The court clearly should have asked more open-ended questions. See Zola, supra, 112 N.J. at 397, 548 A.2d 1022 (approving of open-ended questions to elicit jurors’ feelings about appropriateness of death penalty for murders accompanied by robbery and rape).
Because defendant was charged with robbery, kidnapping, and rape, in addition to capital murder, the trial court simply asked prospective jurors if, assuming conviction on all counts, the existence of those other violent crimes would affect their ability to consider mitigating evidence. The court consistently asked similar minimalist, leading questions about other crimes despite defense counsel’s repeated request for more in-depth voir dire on the subject.
The court’s voir dire regarding the presumption of innocence was also inadequate. In State v. Moore, 122 N.J. 420, 455-56, 585 A.2d 864 (1991), the Court endorsed an individualized inquiry, either through direct questioning or a questionnaire, into juror attitudes about the presumption of innocence. Although the trial court asked a number of specific, open-ended questions on the subject, it allowed problematic juror responses to go unexplored even after defense counsel requested further questioning.9 The court would rehabilitate jurors by engaging in leading descriptions of the presumption of innocence. It failed, however, to follow up by asking further open-ended questions to determine whether the court’s description of the presumption of innocence had educated jurors or whether they still adhered to their problematic responses.
B.
The most fundamental and egregious shortcoming in the jury-qualification process related to the subject of racial bias. Racial *235bias was a critical problem in this case. This case involved the alleged carjacking, rape, and murder of a young, suburban white woman by an urban black man. Statements purportedly made by defendant caused the racial backdrop of the case to be even more focused and prominent. Defendant referred to Huggins as a “white bitch,” there was testimony that he told Tariq Ayres that he had “knocked off some white girl,” and he told Dunn that he would spare a black carjacking victim but kill a white one.10 Because of those racial themes and overtones, defense counsel requested that the trial court engage in extensive voir dire on racial bias and submitted numerous questions to that effect. The proposed questions dealt with racial bias, reaction to the facts of the case, reaction to defendant’s statements, and attitudes about Trenton and the changes that have occurred in cities.
Remarkably, the trial court, stating that race was, at most, “incidental,” that it did not view the facts as “a bias type crime,” and that race “is not the overriding issue by a long shot,” had a different view of the case. The court thus perceived that this case did not implicate race. It stated: “[I]f it were to ask a great number of questions concerning race, the [cjourt would be in effect, by its emphasis, telling the jury that this is a race case. And that is not the circumstance here.” The court betrayed a serious misapprehension about the law that governs the conduct of a prosecution — particularly a capital-murder case — in which race by definition is a patent factor that must be taken into account to assure a fair trial.
The court-conducted voir dire included only minimal questioning about racial bias. Initially, the voir dire simply consisted of some sort of permutation of the following single question: “In this case, the defendant, Mr. Harris, is a black male. The victim, Kristin Huggins, was a 22 year old white female. Given those factors, would that in any way influence your determinations in *236this case?” The question invariably, and predictably, elicited “no” answers. The court asked no other questions about race to each venireperson.
Defendant eventually moved to dismiss the qualified jurors because of the inadequacy of the voir dire concerning racial attitudes; he subsequently moved for a stay of voir dire pending the trial court’s consideration of the motion. The trial court denied the stay, and the Appellate Division affirmed the denial. During oral argument in the Appellate Division, however, Judge Skillman apparently noted, without judging the adequacy of voir dire, that case-specific race questions were permissible if requested by a party. In response to that suggestion, defendant and the State compiled a joint list of ten race questions to be propounded to previously qualified jurors and to future prospective jurors. The trial court accepted the idea of recalling previously qualified jurors, but it refused to ask ten case-specific race questions.
Rather, the trial court’s reformulated voir dire consisted simply of a few leading questions focused on the “white bitch” statement. The trial court generally treated the “white bitch” remark in a vacuum. The Court repeatedly conducted inadequate racial voir dires, including those of jurors who were ultimately seated. Most elicited answers similar to “I don’t think so” in response to the trial court’s question about whether defendant’s allegedly having called Huggins a “white bitch” would affect the potential jurors’ ability to be fair. In each instance, the trial court did not follow up on those uninformative and ambiguous responses. Others expressed a desire to know the context of the remark. For example, one venireperson said: “I’d want to know why he called her that.” The venireperson added: “I don’t think it would affect it. It’s hard to imagine it standing alone. In some context it may, you know, have a different meaning. But by itself, it doesn’t really.” One juror, who was eventually seated, stated: “I don’t think it has a bearing, and I assume that’s the way he talks.” Yet, the trial court failed to ask additional questions to these venireper-sons. Consequently, defense counsel with good reason asserts *237that “the court’s question implied a correct answer and thereby ■virtually assured that the responses would be vapid and hollow, and of no use to counsel.”
The Court has noted that in some eases, a single, general race question can be sufficient to ensure a fair and impartial jury. E.g., State v. Loftin, 146 N.J. 295, 342, 680 A.2d 677 (1996); State v. Ramseur, 106 N.J. 123, 244-48, 524 A.2d 188 (1987). The Court, however, has recognized that a more extensive voir dire is often required:
Racial prejudice may be either blatant and easy to detect or subtle and therefore more difficult to discern. A probing voir dire that elicits more than a “yes” or “no” response will aid the trial court in excusing prospective jurors for cause and will assist the defense in exercising its peremptory challenges. When the defendant is a member of a cognizable minority group, a more searching voir dire should be conducted, if requested.
[Williams II, supra, 113 N.J. at 428, 550 A.2d 1172.]
Several factors determine whether the trial court has an obligation to delve more deeply than one question into prospective jurors’ racial attitudes. These include whether the crime is interracial in nature, Ramseur, supra, 106 N.J. at 245, 524 A.2d 188, whether race is inextricably bound up with the case, id. at 247, 524 A.2d 188, whether the defendant requests specific additional questions, Loftin, supra, 146 N.J. at 342, 680 A.2d 677, and whether defense counsel has an opportunity to delve more deeply into the subject, State v. McDougald, 120 N.J. 523, 551, 577 A.2d 419 (1990).
Under the factors that this Court has enumerated, the trial court’s race-based voir dire was grossly inadequate. In-depth voir dire about race was clearly imperative. First, the crime was interracial in nature. Second, despite the trial court’s firm personal belief to the contrary, race was a central feature of the case given the multiple racially motivated statements attributed to defendant and the crime itself, in which a white suburban woman had gone into the city and gotten carjacked, raped, and murdered by a black man. The murder itself appeared to have been racially motivated. Third, defendant repeatedly requested more extensive voir dire and even submitted an extensive list of proposed questions from which the court could have chosen. Fourth, the court *238terminated all attorney-conducted voir dire, thus precluding deeper inquiry by defense counsel. The shortcomings of defense counsel’s inability to participate fully was further exacerbated by the court’s refusal to go into greater depth with specific jurors when requested by defendant. For example, the comment by a prospective juror who was eventually seated that “I assume that’s the way he talks” and the court’s failure to probe further into that comment are particularly troubling. Fifth, the questions were decontextualized and leading in nature, and thus they elicited short, vague responses. They were not tailored to go beyond obvious bias and to delve into latent bias. A number of jurors appeared to recognize the context-free nature of the questions in their responses, such as “I don’t think so” and “I’d want to know why he called her that.” The court did not follow up on these responses and did not allow counsel to do so.
The majority concludes that “[sjtill, the questioning was sufficiently probing to enable court and counsel to gain a perception of jurors’ attitudes.” Ante at 480, 716 A.2d at 166. I cannot agree. The trial court’s handling of the race issue in voir dire fell far short of this Court’s standards. That error was amplified by the court’s inadequate pretrial-publicity voir dire, especially because of the extensive publicity about defendant’s remarks to an alleged rape victim that she should not resist because he already had killed a white woman.
C.
The purpose of a searching voir dire is to ferret out biases that jurors may not express initially but that they may reveal with more sophisticated, open-ended questions. Voir dire does not exist simply to induce jurors, through leading questions, to abandon problematic responses and to agree with the court’s “correct,” death-qualifying answers so that the jury-selection process can end quickly with minimal dismissals for cause. The inappropriate approach, however, was the one that the trial court chose. The trial court’s voir dire in a number of areas was seriously made-*239quate in its failure to delve substantially into important aspects of the case, its reliance on lectures and leading questions, and its acceptance of problematic responses, which it attempted to cure through leading questions.
The voir dire deficiencies cannot be considered harmless, given the centrality of the deficient aspects of voir dire, defendant’s exhaustion of his peremptory challenges, and the trial court’s refusal to augment the number of defense peremptories.
No defendant should be convicted of capital murder and sentenced to death based on the voir dire that occurred in this case. The Court should reverse defendant’s convictions and vacate his death sentence.
Ill
Early in the proceedings in this case, defendant moved for separate guilt-phase and penalty-phase juries so that he could testify during the guilt phase without exposing his criminal record to the penalty-phase jury. In an accompanying affidavit, defendant stated that he would testify if the trial were bifurcated. Defendant contends that the trial court’s denial of his bifurcation motion effectively deprived him of his right to testify, in violation of his state and federal constitutional rights.
N.J.S.A. 2C:11-3c(1) provides:
Where the defendant has been tried by a jury, the [penalty] proceeding shall be conducted by the judge who presided at the trial and before the jury which determined the defendant’s guilt, except that, for good cause, the court may discharge that jury and conduct the proceeding before a jury empaneled for the purpose of the proceeding.
[ (emphasis added).]
Thus, a trial court may impanel a second jury for the penalty phase if it determines that good cause exists.
In considering the definition of “good cause,” this Court has focused on the existence of highly prejudicial evidence that is only admissible during one phase of a capital trial. See State v. Erazo, 126 N.J. 112, 130-33, 594 A.2d 232 (1991); State v. Dixon, 125 N.J. *240223, 250, 593 A.2d 266 (1991); State v. Long, 119 N.J. 439, 474-75, 575 A.2d 435 (1990). As the Court stated in Dixon, supra: “Given the structured nature of capital sentencing, we have repeatedly emphasized that juries are not to be subjected to extraneous factors that may influence a jury’s verdict in a way neither contemplated nor authorized by statute.” 125 N.J. at 250, 593 A.2d 266.
The Court has recognized that, at times, a limiting instruction is insufficient and that limited admissibility of highly prejudicial evidence constitutes good cause, thus requiring a bifurcated jury. As the Court stated in Erazo, supra: “A separate penalty-phase jury commends itself when guilt-phase evidence is so prejudicial that the same jury could not fairly sit on both phases of the trial.” 126 N.J. at 133, 594 A.2d 232. In Biegenwald IV, supra, 126 N.J. at 44-45, 594 A.2d 172, the Court held that reliance by the State on the prior-murder aggravating factor during the penalty phase generally would require separate juries because of the inadmissibility of that evidence during the guilt phase and the necessity of discussing that aggravating factor at voir dire.
The trial court concluded that the concerns expressed in Erazo and other like cases had been “assuaged” by State v. Brunson, 132 N.J. 377, 625 A.2d 1085 (1993), which required trial courts to “sanitize” criminal convictions that are presented to the jury. In fact, the trial court implied that Erazo’s endorsement of jury bifurcation in this type of ease was no longer good law because Brunson effectively overruled it. That conclusion, however, was erroneous.
Under those standards, this case presents good cause. Defendant had an extensive record of serious and violent offenses that extend back a significant period of time. Had this record been introduced even in sanitized form during the guilt phase, a limiting instruction would have been insufficient to prevent it from substantially affecting the jury’s deliberations during the highly subjective penalty phase. In light of the indication by six of the twelve seated jurors that they considered a defendant’s record at *241least somewhat relevant in determining the appropriateness of a given sentence, defendant stresses the high probability of such a prejudicial impact.
Another reason why jury bifurcation would have been highly appropriate in this ease is that, unlike the other cases in which-this Court and lower courts have considered jury-bifurcation motions, this case involved limited-admissibility evidence that the State could introduce only if defendant were to testify, in which case defendant’s record would have impeached his credibility. Before trial, defendant indicated by affidavit that he wished to testify but that the lack of jury bifurcation would prevent him from doing so because of the prejudicial impact of his criminal record. -As a result of the trial court’s refusal to bifurcate the jury, defendant elected not to testify in his own defense. Consequently, the trial court’s decision not to bifurcate the jury must be evaluated not only in light of the prejudicial impact of defendant’s record on the jury’s penalty-phase deliberations but also in light of defendant’s right to testify in his defense, see Rock v. Arkansas, 483 U.S. 44, 107 S.Ct. 2704, 97 L. Ed.2d 37 (1987) (holding right to testify was federal constitutional right); State v. Savage, 120 N.J. 594, 626-28, 577 A.2d 455 (1990) (same holding under New Jersey Constitution).
In the context of a capital prosecution, a violation of the right to testify cannot be deemed harmless, because of its fundamental nature. State v. Rosillo, 281 N.W.2d 877, 879 (Minn.1979) (holding violation of right to testify could not be harmless). Further, defendant was prejudiced by the trial court’s refusal to bifurcate. It is safe to assume that he would have challenged Gloria Dunn’s allegation that he was the triggerman and that he would have testified that she actually pulled the trigger. That may be inferred from the defense strategy generally, which mostly focused on Dunn, and on defendant’s allocution at his noncapital sentencing proceeding, in which he insisted that Dunn was the murderer. Thus, assuming that the court had granted the bifurcation motion and defendant had taken the stand and testified that Dunn, not he, *242was the triggerperson, defendant’s testimony reasonably could have affected the verdict.
Therefore, I conclude that the trial court’s erroneous refusal to bifurcate the trial was not harmless beyond a reasonable doubt. The Court should vacate defendant’s death sentence based on the reasonable possibility that defendant’s testimony could have affected the determination of defendant’s status as triggerman and, thus, defendant’s death-eligibility.
IV
Relying on its guilt-phase presentation, the State presented no new evidence during the penalty phase. The instructions that the jury received to guide its consideration, application, and evaluation of that evidence in reaching the ultimate determination — whether defendant should be put to death — were grossly inadequate, contradictory, and confusing.
Before defendant began his penalty-phase presentation, the trial court instructed the jury that “what the state has effectively done is this, it has placed in evidence all that you heard during the guilt phase of the trial, and it relies upon that to establish the aggravating factors which it alleges.” In its opening instructions to the jury, the court had stated that “the only evidence which you may consider that was presented during the guilt phase of the trial is that evidence which may be used to support one or both of the aggravating factors alleged by the state.” In its closing charge to the jury following the penalty-phase presentation, the court, after defining the elements of the felony-murder aggravated factor, stated that
the guilt and sentencing phases of the trial are considered separate proceedings. Therefore, you must deliberate anew, regarding any facts established in the guilt phase which the state relies upon to prove an aggravating factor. And you have a right to reach a different conclusion about such facts in the penalty phase.
Despite those instructions, the court never guided the jury about specific segments of guilt-phase evidence that may not have been admissible in the penalty phase. Although defendant did not *243request such an instruction, its omission by the court constitutes a breach of its duty to ensure a fair trial and amounts to plain error.
Because the only issues before the jury during the penalty phase are the determination of aggravating and mitigating factors and the balancing of those factors, evidence admissible at the guilt phase is not necessarily admissible at the penalty phase. Dixon, supra, 125 N.J. at 249-50, 593 A.2d 266; State v. Rose, 112 N.J. 454, 507-08, 548 A.2d 1058 (1988). Because of the distinction between the two phases, the trial court must instruct the jury about guilt-phase evidence that the jury may not consider during its penalty deliberations. As the Court stated in Erazo, supra:
When the same jury hears both phases of the trial, the court should provide instructions on the extent to which the jury may use guilt-phase evidence [in] its penalty-phase deliberations. Even when guilt-phase evidence is not incorporated in the penalty phase, the danger abides that the jury will rely on it during the penalty-phase deliberations. Thus, the court should instruct the jury concerning the evidence that it may use in its penalty deliberations and the purposes for which that evidence may be used.
[126 N.J. at 133, 594 A.2d 232 (citation omitted).]
Several aspects of the guilt phase were not admissible during the penalty phase. Gloria Dunn’s testimony regarding defendant’s offer to “pop” her ex-boyfriend was admissible during the guilt phase, but was not admissible to prove either of the aggravating factors. Defendant’s racially-based statements, including his references to Huggins as “white bitch,” his purported statements that he had “knocked off some white girl,” and his alleged assertion to Dunn that he would kill a white victim but spare a black victim, assuming they were admissible during the guilt phase, were clearly inadmissible to prove either of the aggravating factors.
The trial court, however, instructed the jury only that it had to redeliberate on any guilt-phase evidence on which it relied during the penalty phase and that the jury could rely only on guilt-phase evidence that was relevant to the aggravating and mitigating factors alleged. The trial court erred in not specifying which aspects of guilt-phase evidence could be used during the penalty phase and which could not. Had the court instructed the jury to *244disregard that evidence at the penalty phase, the highly subjective balancing process that resulted in defendant’s death sentence may have been different. The error prejudiced defendant. The Court should vacate defendant’s death sentence.
V
Defendant twice requested that the trial court charge the jury that it could find defendant guilty of the crime of murder under N.J.S.A. 2C:11-3a without being unanimous about whether he was guilty of purposeful-or-knowing murder or felony murder. The trial court denied the request. Instead, the trial court submitted the two types of murder separately. The court instructed the jury that “[bjecause the elements or the criteria of purposeful or knowing murder are different from those of felony murder, you must consider each separately. You may find the defendant not guilty of both, guilty of both, or guilty of one but not of the other.”
Although the Court upheld a similar jury instruction in State v. Cooper, 151 N.J. 326, 356-63, 700 A.2d 306 (1997), given the dynamics of a capital murder prosecution and the extraordinary importance of unanimity as the indispensable basis for determining both death-eligibility and deathworthiness, I believe it is patently unfair not to require the felony-murder nonunanimity instruction, id. at 431-42, 700 A.2d 306 (Handler, J., dissenting). A nonunanimity instruction on an element of murder is an indispensable condition of defendant’s death-eligibility. It goes to the heart of the due process and punitive requirements of the death penalty and consequently is an essential part of the ultimate fundamental fairness that must surround capital prosecutions. Accordingly, I continue to insist that such an instruction be given.
VI
In response to a defense request to inform the penalty-phase jury of what the aggregate sentence likely would be if defendant were spared the death penalty, the trial court determined that it instead would inform the jury of the minimum and maximum *245sentence for each noncapital conviction and of its obligation not to consider those sentences in deciding whether to sentence defendant to death. The court told counsel that “[w]hat the [c]ourt has not done is it has not included a summary paragraph, suggested by the defense, explaining the maximum number of years that could be imposed and the maximum amount of parole ineligibility. The reason the Court has not done that is that it does not know that it would impose or could even rationally impose that many years.”
In its instruction, the court informed the jury of the minimum and maximum term for each offense and the highest possible parole bar. The court then stated:
The [c]ourt will decide at a separate sentencing proceeding whether those sentences will be consecutive or concurrent to the sentence to be imposed for purposeful-or-knowing murder. Consecutive sentences merely being one follows the other, and there is a continuum and you add the number. Concurrent meaning that they are all to be served at one time.
The possible sentences for the other convictions should not influence your decision regarding the appropriateness of a death sentence on the murder charge. . Your decision must be based only upon aggravating and mitigating factors presented by the evidence.
Despite having professed that “it does not know that it would impose” “the maximum number of years that could be imposed and the maximum amount of parole ineligibility,” the court proceeded to mete out precisely those maximum sentences. After the death verdict, the trial court imposed the maximum sentence, including an extended term of life imprisonment with twenty-five years of parole ineligibility for aggravated sexual assault, on each nonmerged count and ordered that they run consecutively to each other and to the sentence that defendant had been serving. If defendant’s death sentence were to be vacated, his prison sentence for the crimes against Huggins would total two life terms plus fifty-five years with eighty-two and one half years of parole ineligibility. That sentence would be consecutive to his prior sentence of life imprisonment with a thirty-year parole bar. Hence, if defendant’s death sentence were reversed, he would face a total of 112/6 years of parole ineligibility. Obviously, defendant, *246who was forty at the time of his arrest, would die in prison if he were not executed.
The trial court should have informed the jury that defendant would receive the maximum aggregate sentence on the noncapital convictions. The majority’s rationale for keeping the jury uninformed is totally unsatisfactory and terribly unfair to a defendant facing a death sentence.11
This Court repeatedly has held that penalty-phase juries must be informed of the legal effect of their sentencing decisions, including the noncapital sentences to which the defendant is subject. Consequently, the Court has required that juries be apprised of defendants’ prior sentences, State v. Bey, 129 N.J. 557, 603, 610 A.2d 814 (1992) (Bey III), cert. denied, 513 U.S. 1164, 115 S.Ct. 1131, 130 L. Ed.2d 1093 (1995), and of the possible noncapital sentences resulting from the current capital prosecution, Martini I, supra, 131 N.J. at 313, 619 A.2d 1208.
In Loftin, supra, the Court held that
in future eases, if the court, based on the evidence presented believes that there is a realistic likelihood that it will impose a sentence to be served consecutively to any of defendant’s prior sentences, in the event the jury does not return a death sentence, the jury should be so informed. We believe that in most cases the courts will conclude that there is a “realistic likelihood” that it will impose a consecutive sentence rather than a concurrent sentence in the event of a non-death verdict. However, not every court necessarily will reach that conclusion. In those cases, the court need not inform the jury whether a non-death sentence is likely to be consecutive or concurrent.
[146 N.J. at 372, 680 A.2d 677.]
The rationale in Loftin was that, in order for the jury to be fully informed about the results of its various sentencing options, it *247should know whether the sentences are likely to run consecutively. Though Loftin concerned whether the noncapital sentence would run consecutively to a prior sentence, that rationale applies to the contemporaneous-offense context. Although the trial court may not be sure what the exact sentences will be for the noncapital offenses, it may often — and, indeed, usually, if not invariably— have an idea about whether it will impose those sentences consecutively to the murder sentence.
The trial court should have told the jury of the possible sentences that defendant faced on each noncapital count and the likelihood that those sentences will run consecutively to the murder count. That would have apprised the jury of the likely legal effect of its decision.
Though the trial court complied with the first requirement by informing the jury of the possible sentences, it failed to tell thé jury the likelihood that it would impose consecutive sentences. It justified that decision by stating that it did not know if it “rationally” could impose the sentences consecutively. Given the trial court’s eventual imposition of maximum consecutive sentences, including an extended term for aggravated sexual assault — and its pointed and dramatic post-penalty-phase verdict statement that, had it been a juror, the court would have been the first to vote to execute defendant — the court’s reasons for not so advising the jury appear disingenuous and are very unconvincing. It is naive for this Court to accept the trial court’s protestation that it was unsure of the likelihood it would impose consecutive sentences.
The imposition of consecutive sentences in this case follows a pattern of trial courts imposing consecutive sentences in capital cases. In an example similar to this case, the trial court in Martini I, supra, 131 N.J. at 207, 619 A.2d 1208, imposed a consecutive extended term of a life sentence with a twenty-five year parole disqualifier on the defendant’s kidnapping conviction. In Loftin, supra, 146 N.J. at 333, 680 A.2d 677, the trial court imposed the defendant’s capital sentence consecutively to the life sentence he had received for committing a prior murder. Because *248in this ease the trial court imposed the maximum aggregate sentence against defendant and, in open court, expressed its subjective opinion that defendant should be executed and because courts invariably impose consecutive sentences in capital cases, the court was virtually certain that it would impose consecutive sentences on defendant. It should have informed the jury of its intention.
The majority, finding no error in the trial court’s instructions, writes:
The trial court followed the Martini guidelines. It explained the sentencing options for each of the offenses (omitting reference to possible extended terms for repeat offenders), that he would decide later whether to impose those sentences concurrently or consecutively, and that the potential sentences on other convictions should not affect their determinations on life or death.
[Ante at 197, 716 A.2d at 496.]
While Martini I merely required that the court inform the jury of the possible sentences and that the decision about the actual sentences and whether those sentences will be consecutive is for the court and is not to be considered, the rationale of Loftin requires that the trial court tell the jury that it will impose consecutive sentences when there is a reasonable likelihood that it will do so. But, the trial court did not inform the jury that it would impose consecutive sentences on defendant.
The trial court’s omission cannot be considered harmless. See Loftin, supra, 146 N.J. at 431, 680 A.2d 677 (Handler, J., dissenting) (concluding trial court’s failure to inform jury of defendant’s likely parole ineligibility was reversible error). Although, in his penalty-phase summation, defense counsel repeatedly told the jury that its choice was between lethal injection and “incarcer[ation] for the rest of his life,” that reference was to the judge’s instruction that defendant would receive between thirty years and life for the murder conviction. Counsel never argued to the jury that defendant would receive consecutive sentences for the noncapital counts. The Court “agree[s] that most trial judges, considering the type of crimes of which defendant was convicted, would find a ‘realistic likelihood’ that some of the non-death sentences would be *249made consecutive to the 30-year parole disqualifier for murder.” Ante at 198, 716 A.2d at 496. Nevertheless, it finds “that defendant’s jury could not have misunderstood the gravity of the non-capital sentences defendant would have faced.” Ante at 198, 716 A.2d at 496. We cannot in this context guess about what the jury understood. Neither counsel’s summation nor the trial court’s instructions cure the prejudice that defendant suffered from the absence of a clear instruction by the trial court to indicate that defendant would spend the rest of his life in prison if not sentenced to death.
Moreover,the court erred by instructing the jury to disregard defendant’s noncapital sentences. See State v. Nelson, 155 N.J. 487, 504-06, 715 A.2d 281, 290 (1998). The error was not harmless. See id. at 504, 715 A.2d at 289-90 (Handler, J., dissenting).
This Court should reverse defendant’s death sentence due to the trial court’s failure to inform the jury that it would likely impose consecutive noncapital sentences on defendant and give defendant no prospect of ever leaving prison alive. The trial court’s erroneously instructing the jury to disregard defendant’s noncapital sentence also mandates reversal.
VII
Significant and irremediable errors infected this prosecution and tainted defendant’s convictions and death sentence. The court’s muted response to the pretrial and midtrial publicity is the gravest of the errors.
Members of the venire were exposed to varying degrees of vicious, inflammatory pretrial publicity, which affected many critical features of the case, such as the cross-racial nature of the crime and the fact that the crime allegedly had occurred along with a kidnapping, robbery, and rape; further, it implicated important concerns such as the presumption of innocence, the ultimate guilt of defendant, and the justification for the death sentence. The trial court took minimal steps to cure any prejudice arising from the pretrial publicity. Despite the ubiquity of *250the publicity in Mercer County, the court refused to change venue and merely impaneled a jury from Burlington County. The court’s voir dire regarding pretrial publicity was far too cursory to uncover the injurious effects the publicity had on the jury venire.
In addition to the trial court’s failure to change venue and to conduct a full and searching voir dire on exposure to pretrial publicity, the court utterly failed take serious measures to counteract exposure to midtrial publicity, which included an unsubstantiated allegation that defendant had committed a prior murder. Defendant’s exposure to intense and prejudicial midtrial publicity thwarted a trial by an impartial jury. In failing to remediate that publicity, the court deprived defendant of an unbiased jury. Defendant had a right to be judged by an impartial jury guided solely by the evidence arising within the confines of the courtroom. This right was seriously impaired by a combination of the Trento-nian’s death crusade and the trial court’s unwillingness to counteract that crusade even after recognizing its existence. That confluence of circumstances denied defendant a fair trial and impugns the very foundation of his convictions and death sentence.
Further, the trial court’s death-qualification voir dire had substantial shortcomings. The court repeatedly, through highly leading questions, coached jurors into agreeing that they could follow the law. The trial court often, without delving further into what the jurors really believed, allowed jurors to make wildly contradictory statements about whether the death penalty should be automatic for premeditated murders. The court almost totally shut out counsel. Except for the voir dire of the first seven venireper-sons, counsel was not allowed to ask questions directly. Moreover, even if counsel informed the court of specific problematic responses by a venireperson, the court almost never asked further questions in response to those concerns. In general, the only questions suggested by counsel that the court then asked were the court’s standard questions that it had forgotten to ask. This feeble voir dire was incapable of ensuring that the jurors who sentenced defendant to die were qualified to sit on a capital jury. *251The inadequate death-qualification was exacerbated by the court’s failure to voir dire the venirepersons sufficiently in respect of the racial overtones that pervaded the case. The court did not appreciate the pervasive role of race in this case and, accordingly, conducted a shallow voir dire that was unable to root out venire-persons’ overt or subtle racial prejudices. The trial court’s deficient voir dire requires reversal of defendant’s convictions and death sentence.
This capital case should have been tried by a bifurcated jury. Given the seriousness and length of defendant’s record, the similarity of the record to some of the offenses alleged at trial, defendant’s indication that he wished to testify if his record were shielded from the penalty-phase jury, and the intense and subjective nature of the penalty phase, the trial court’s decision not to bifurcate the jury was clearly an abuse of discretion. The error was patent in view of the trial court’s determination that this Court’s decision in Erazo was no longer valid and could be disregarded. As noted, this Court has stated repeatedly that bifurcation is favored in precisely this type of case. When confronted with compelling circumstances for bifurcation, however, the trial court misapplied the law by determining that Erazo was no longer controlling authority. That ruling unduly burdened defendant’s right to testify. The court’s misinterpretation confronted defendant with an unacceptable choice of testifying, and hence poisoning the jury’s penalty-phase deliberations, or not testifying, and hence not rebutting the State’s case. Because defendant’s testimony would have cast doubt on his death-eligibility, the trial court’s refusal to bifurcate the jury requires reversal of his death sentence.
Inadmissible and prejudicial guilt-phase evidence was allowed to pour over into the penalty phase. Consequently, the absence of penalty-phase instructions informing the jury that specific guilt-phase evidence was inadmissible in the penalty phase mandates reversal of defendant’s sentence. Finally, the court’s failure to charge that it intended to impose consecutive sentences on defen*252dant for the noncapital convictions, a sentence that would have forced defendant to spend the rest of his life in prison, requires reversal of the death sentence.
This decision exemplifies the inconsistency, chronic contradictions, and confusion that surround and permeate our capital-murder jurisprudence. Worse, the decision illustrates the regression in capital-punishment jurisprudence. Time and again, the Court has fashioned and pronounced sound and salutary principles to structure and guide the prosecution of capital-murder cases. Those principles reflect the singular protections that are required when a defendant is on trial for his life. They are the indispensable basis for the assurance, assuming such assurance is humanly possible, that the trial for capital murder is sufficiently fair under constitutional standards and that it will generate that level of comfort and confidence that society demands and ejects before it will sentence a criminal to death. Yet, time and again, in successive cases, the Court repeats but nonetheless retreats from those principles by refusing to apply them in compelling situations that are often more egregious than those giving rise to the promulgation of the principles in the first place.
Perhaps this inconstant jurisprudence mirrors the intense ambivalence that is engendered by capital punishment and the controversy that pervades its administration. But the Court, in our constitutional judicial structure, is the institution responsible for assuring that the prosecution of capital eases will comport with constitutional standards and for preserving the legal regime envisioned by our Constitution. The Court cannot allow itself to be conflicted by the legal and social dilemmas that surround capital punishment. It must be resolute, consistent, and firm in its disposition of capital causes. Regrettably, in this case, it wavers.
For affirmance — Chief Justice PORITZ and Justices POLLOCK, O’HERN and GARIBALDI — 4.
*253For affirmance in part: reversal in part — Justices STEIN and COLEMAN — 2.
For reversal and vacating — Justice HANDLER — 1.
Defendant submitted a five-volume appendix to this Court that contains hundreds of articles chronicling the extent and nature of the deleterious publicity, both before and during trial, that surrounded this case and that were carried in the Trentonian, the Trenton Times and other newspapers.
Most publicity by far occurred between the murder and April 1993, when defendant was arrested. It again spiked when defendant and Gloria Dunn were indicted in mid-1993, when defendant dismissed his attorneys at the end of 1993, when defendant was tried and convicted of an unrelated robbery charge in mid-1994, when Gloria Dunn pleaded guilty and agreed to testify for the State in late 1994, and as the trial of the present case began and proceeded in late 1994 and early 1995.
The Trentonian ran headlines stating "Artist in hearts of many,” "Thoughts of art filled her head hours before death,” “Dear G-d, take good care of Kristin," "Kristin Huggins' death swept the residents of Trenton with waves of grief,” and “City grieves year after artist’s brutal murder.” The Trenton Times ran a front-page headline stating "Missing artist's family lives in pain-filled limbo.” When Huggins's parents testified at trial, the Trentonian further emphasized its sympathy.
This headline referred to defendant's alleged terrorization of other inmates.
This headline referred to defendant's attempt to change his legal name to Konko Lumumba.
The Courier Times, a Pennsylvania newspaper, ran an article entitled "Huggins Suspect 'Would Kill You in a Heartbeat.’ ”
The nickname "Squirt Boy" refers to the lethal injection the Trentonian has hoped defendant will receive.
This article contained the following set-aside quote: "Most people figured the jury would think, ‘We’ll have lunch on the county, and we'II squirt him — this afternoon.'"
The Court also cited favorably to the American Bar Association Standards for Criminal Justice, Standards Relating to Fair Trial and Free Press § 8-3.5(f) (1978), which states:
If it is determined that material disseminated during the trial goes beyond the record on which the case is to be submitted to the jury and raises serious questions of possible prejudice, the court may on its own motion or shall on motion of either party question each juror, out of the presence of the others, about his exposure to that material.
[Bey I, supra, 112 N.J. at 87-88 & n. 28, 548 A.2d 846 (internal quotations omitted) (emphasis added).]
One juror stated that if the defendant did not testify, he must be "comfortable” with the State's story. The court did not attempt to elicit even the slightest explanation and did not return to the subject despite defense counsel’s specific request that it do so.
Dunn's testimony regarding the role of Huggins's race in defendant's intent to kill was a surprise and, thus, was unknown during voir dire.
In my opinion, under Simmons v. South Carolina, 512 U.S. 154, 114 S.Ct. 2187, 129 L. Ed.2d 133 (1994), defendant, by telling the jury how much time he was likely to serve as a result of committing those felonies in addition to the murder, was entitled to rebut the State’s allegation that he deserved the death penalty because he committed the murder in the course of various felonies. Cf. Loftin, supra, 146 N.J. at 425, 680 A.2d 677 (Handler, J., dissenting) (concluding parole ineligibility should be admissible as mitigating evidence to rebut prior murder aggravating factor).