Because the majority resorts to an improper application of the harmless error doctrine where County Court committed a flagrant Molineux violation in allowing highly prejudicial evidence as to defendant’s prior federal felony conviction and a multitude of other collateral evidence concerning prior criminality, I respectfully dissent.* The glaring error of permitting an FBI special agent to testify at length as *469to defendant’s prior cargo theft in April 2000—a crime of the very same nature as the charges in this case—most certainly impacted the jury to defendant’s detriment. In addition, as this prosecution rested entirely on circumstantial evidence of less than compelling force, invocation of harmless error is not appropriate to avoid the conclusion that defendant was deprived of a fair trial.
Furthermore, County Court’s admission of several hundred pages of testimony on numerous other collateral matters is most troubling. In addition to the Molineux error in allowing evidence of defendant’s prior conviction for cargo theft, the court allowed evidence as to the underlying facts of defendant’s alleged involvement in an uncharged cargo theft with an accomplice, one Nelson Quintanilla, in 1996, in yet another tractor-trailer heist. These errors were then further compounded by the admission of testimony from police officers as to the felony conviction of one Jose Gotay—a person not charged in this case—for receiving stolen cargo in December 2000 and his involvement in another such offense in December 2003. All of these collateral matters involved similar types of offenses as those with which defendant here was charged.
According to the majority, there was a nexus bonding the events in this case with Gotay’s criminal past based upon phone records showing that calls from one of defendant’s cell phones were made at about the time of the crime to a company where Gotay was listed as a contact person, which “supported an inference that defendant at that moment needed a fence’s services” (see majority op at 465) sufficient to inject evidence of this third party’s crimes into defendant’s trial. As to the evidence involving Quintanilla, the majority states that “the evidence supported a finding that Quintanilla and the defendant were working together to commit the crime in this case” (see majority op at 466) when he allegedly abandoned the trailer in northern New Jersey. The majority concludes that the admission of these bad acts by other persons, where it is alleged that defendant was merely in contact with them at or about the time of the commission of the crime, was not unduly prejudicial to defendant. I disagree.
The prejudicial effect that all of this extensive collateral evidence had is clear: the jury was naturally led to believe that, because defendant had committed two cargo thefts in the past *470and at the time of this alleged crime telephoned a federally convicted “fence” of stolen goods, he was guilty of the crimes charged here. It is remarkable that one fifth of this trial record—spanning several hundred pages of trial testimony from numerous state and federal law enforcement officers and dozens of exhibits—is consumed by collateral matters, including multiple photographs depicting warehouses, stolen trucks and stolen goods related to extraneous crimes. In total, two FBI agents and three police officers testified as to uncharged matters and convictions involving cargo thefts and stolen goods. Much of the prosecution’s summation was then centered upon these matters.
Our long history of more than a century of Molineux jurisprudence guards against the admissibility of a defendant’s prior crimes, except in very limited circumstances, because of the real danger that such evidence would sway the jury to convict the defendant based upon prior bad acts and thus deprive the accused of the fundamental right to a fair trial (see People v Molineux, 168 NY 264, 313 [1901]). To introduce Molineux evidence, the prosecution must meet its burden of establishing that the probative value of the proffered material outweighs its prejudicial effect (see People v Alvino, 71 NY2d 233, 241-242 [1987]).
“Prejudice involves both the nature of the crime, for the more heinous the uncharged crime, the more likely that jurors will be swayed by it, and the difficulty faced by the defendant in seeking to rebut the inference which the uncharged crime [or prior charged crimes] evidence brings into play” (People v Robinson, 68 NY2d 541, 549 [1986]).
The reason for this is obvious: “ ‘it is much easier to believe in the guilt of an accused person when it is known or suspected that he [or she] has previously committed a similar crime’ ” (People v Allweiss, 48 NY2d 40, 48 [1979], quoting People v Molineux, 168 NY at 313). Hence, the admission of such collateral evidence must be subjected to “the most rigid scrutiny” (Molineux, 168 NY at 313).
I agree with the majority’s holding that County Court erred in allowing evidence of defendant’s prior federal conviction, but this error cannot be characterized as harmless. As the majority notes, the requirements for the Molineux identity exception were not established as to this past crime. Defendant’s prior conviction for a cargo theft was not so unique or peculiar that it *471involved a signature offense. As a prosecution witness explained at the trial, the theft of a tractor-trailer is a relatively common type of crime in this general region. Further, no witness at this trial could testify as to how defendant may have stolen this vehicle, much less that he did so in a peculiar and distinctive manner, thereby identifying himself as the culprit. Nor was the collateral evidence probative of a common scheme or plan because that crime and the charges here do not strike “ ‘such a concurrence of common features that the various acts are naturally to be explained as caused by a general plan of which they are the individual manifestations’ ” (People v Fiore, 34 NY2d 81, 85 [1974], quoting 2 Wigmore, Evidence § 304, at 202 [3d ed]). Moreover, even had the identity or common scheme or plan exceptions been established—which they were not—the probative value of this collateral evidence would have been outweighed by its potential for prejudice.
As to the evidence of defendant’s uncharged cargo theft with Quintanilla, the majority concludes that this is classic Molineux identity evidence. They point out that previously defendant was accused of acting with several individuals in a cargo theft, one of whom he allegedly telephoned at the time of this crime. This mere alleged similarity, however, is not enough to meet the strictures of our Molineux jurisprudence.
To establish the identity exception, Molineux demands that there be evidence of a distinctive crime forming a uniqueness that necessarily identifies the defendant as its culprit, as would a signature. “A similar modus operandi is not enough; there must be sufficient uniqueness to identify the accused as the perpetrator of the crime charged” (Prince, Richardson on Evidence § 4-514 [Farrell 11th ed], citing People v Robinson, 68 NY2d at 548-549; People v Beam, 57 NY2d 241, 251-252 [1982]; People v Allweiss, 48 NY2d at 47-48). As Molineux long ago stated, “ ‘the naked similarity of . . . crimes proves nothing’ ” (Robinson, 68 NY2d at 549, quoting Molineux, 168 NY at 316).
Here, this uncharged cargo theft displays neither any sort of distinctive manner that conclusively identifies defendant in the allegations in this case, nor a pattern of criminal behavior. Rather, this evidence accomplishes nothing more than telling the jury of defendant’s propensity to commit this type of offense based upon his alleged earlier engagement in a similar type of crime along with a person he may have telephoned at the time of this crime. Unknown are the contents of the telephone calls made from defendant’s cell phone to Quintanilla or evidence *472sufficient to substantiate the pure supposition that he was involved in the planning and execution of this crime. Thus, the purported linkage to Quintanilla, upon which extensive collateral evidence was based, is speculative and tenuous. Meanwhile, the potential for prejudice injected into the trial by such extensive collateral evidence is clear. As the prosecution argued to the jury in summation:
“Maybe they are friends. It’s good to be friends with your colleagues. That doesn’t mean that at times they’re not involved in some wrongdoing, and this—on this particular date, the calls surrounding this July 26, 2003, incident, I submit to you the evidence is they’re up to no good, they’re scheming about stealing this trailer, and that’s what happened . . .
“This is some planning to steal the trailer here and then sell the contents and make some money from it.”
It is settled law that when evidence is of slight value when compared to the possible prejudice to the accused that it must be excluded (see People v Allweiss, 48 NY2d at 47).
County Court then added to this prejudicial effect by allowing evidence of Gotay’s conviction and criminal involvement related to the receiving of stolen goods. The court concluded that Gotay, who remains uncharged in this case, was going to act as a “fence” for these goods merely because defendant had telephoned him at the time of the crime, and that this bare fact was sufficient to put such explosive evidence before the jury. Again, there was no evidence of what the defendant and Gotay may have said in these phone conversations. In considering the admissibility of evidence of prior crimes of third parties as part of the prosecution’s case, though not strictly subject to Molineux, the court must be mindful of similar dangers, such as instances where prejudice outweighs probativeness and the danger of the defendant’s propensity to commit the charged crime and his guilt thereof merely by association with individuals with criminal pasts.
We have not hesitated to reverse convictions affected by admission of evidence of uncharged crimes, where the prejudice to defendant was much less acute than in this case (see People v Resek, 3 NY3d 385, 387 [2004] [evidence that police were monitoring a car based upon a report that it was stolen could have *473been dealt with by less prejudicial means, such as instructing jurors that the defendant’s stop of the vehicle and arrest were lawful]; People v Green, 35 NY2d 437, 442-443 [1974] [the trial court erred by allowing police testimony that one month earlier they had attempted to enter defendant’s apartment because of a drug complaint and the error was prejudicial]). In the recent case of People v Giles (11 NY3d 495, 500 [2008]), we held that evidence of prior crimes to show that a debit card was stolen was irrelevant, where there was no proof that defendant committed the prior burglaries, and the resulting prejudice required reversal of defendant’s conviction in the absence of a limiting instruction by the trial court.
The majority ascribes no prejudicial effect from multiple law enforcement agents testifying as to these third parties’ cargo thefts on the basis that this extensive testimony does not directly relate to defendant’s criminal past. They state with confidence that the jury did not impute criminal propensity to defendant from Gotay’s conviction and criminal involvement. The jury, however, would naturally have believed that, because defendant associated with a person convicted of receiving stolen goods in the recent past and was in direct communication with him at the time of this cargo theft, defendant committed the crimes charged.
Turning to harmless error, this doctrine is applicable only if two discrete factors are clear: (1) the quantum and nature of the evidence against the defendant must be great enough to excise the error, and (2) the causal effect that the error may nevertheless have had on the jury must be overcome (see People v Crimmins, 36 NY2d 230, 240 [1975]). Put differently, it must be established that the evidence against the defendant is overwhelming, such that it is likely that the trial error did not infect the jury’s finding (see id. at 240-242).
Here, the majority contends that the noncollateral evidence against defendant was so compelling and overwhelming that it “exclude[s] to a virtual certainty any hypothesis of defendant’s innocence” (see majority op at 468). The noncollateral evidence, however, was neither overwhelming nor particularly compelling. The prosecution’s fingerprint evidence taken from a New York State Thruway toll ticket consisted of a partial, smudged left index fingerprint (allegedly from defendant) sharing a loop that is common to 70% of the population. The prosecution urged that this toll ticket handed to a toll booth collector on the route that the driver may have taken was from the only five-axle *474truck driving that portion of the highway at around the time of the crime. The second piece of evidence was cell-site information showing that calls were made from one of defendant’s cell phones that ostensibly track the same route. There was no testimony from eyewitnesses, DNA evidence, inculpatory statements by defendant, or any contraband (cash or merchandise) recovered from defendant or anyone else. This evidence was certainly not overwhelming in establishing defendant’s guilt.
The prejudice to defendant, however, in allowing an FBI agent to testify as to his prior federal felony conviction, along with the underlying facts of that crime, and the admission of another federal agent’s testimony regarding the facts from a previous tractor-trailer offense, is significant. Allowing the evidence that defendant contacted a federally convicted “fence” compounded the prejudicial effect of these errors. Finally, the fact that at least one fifth of this trial was dedicated to collateral matters casts serious doubts on whether defendant received a fair trial.
Nor were the errors in admitting extensive testimony of extraneous criminal acts cured by the limiting instructions. County Court’s jury instruction first contains a summary of the collateral evidence against defendant, merely highlighting this wrongful evidence to the jury, and then the court told the jury that it must not view these crimes for propensity but for a common scheme or plan and as identity evidence. While the instruction may have served to highlight the wrongly admitted evidence, it certainly failed to cure the prejudice to defendant.
In conclusion, the admission of evidence regarding defendant’s prior bad acts and the bad acts of third parties was highly prejudicial and served to deprive defendant of a fair trial. Where the evidence is far from overwhelming, it cannot be said that the result would have been the same if it were possible to extricate such egregious errors.
Accordingly, I would reverse defendant’s judgment of conviction and direct a new trial.
Judges Graffeo, Read and Pigott concur with Judge Smith; Judge Ciparick dissents and votes to reverse in a separate opinion in which Chief Judge Lippman and Judge Jones concur.
Order affirmed.
It is true, as the majority holds, that there is no unique or specialized crime exception under People v Molineux (168 NY 264 [1901]) providing for the admission of evidence of a prior complex crime to prove the identity of a defendant for the crime charged. If that were so, a jury would naturally infer *469that the defendant was guilty of the instant crime based solely upon propensity, and the ensuing prejudice would be almost insurmountable.