Upon remittitur from the Court of Appeals (12 NY3d 342 [2002]), judgment, Supreme Court, New York County (Harold B. Beeler, J.), entered June 29, 2006, reversed, on the facts, without costs, and the matter remanded for a new trial.
Saxe and Acosta, JJ., concur in a separate memorandum by Saxe, J., as follows: On this appeal we are required to consider the evidence in a case where a man died under circumstances that led the trial court to rule that he committed suicide. On our first review of that determination, we held, by a vote of 3-2, that as a matter of law, the common-law presumption against suicide had not been sufficiently rebutted (48 AD3d 37 [2007]). An appeal to the Court of Appeals followed. The Court of Appeals disagreed with our reliance on the presumption to determine the appeal as a matter of law, observing that “the evidence was strong enough to permit a finding of suicide, though not to require it,” and remitted the matter to this Court for exercise of our weight of the evidence review power (12 NY3d 342, 347 [2009]). Following the Court of Appeals’ instructions, and conducting a weight of the evidence review, a plurality of this Court now concludes that while there was evidence that permitted a finding of suicide (see id.), it was not strong enough to outweigh the evidence tending to point to death by means other than suicide, and that therefore a new trial is needed. A third justice concurs with the conclusion that a new trial is necessary, but declines to reach the weight of the evidence issue, concluding instead that the erroneous mid-trial ruling allowing defendant to present expert testimony alone requires a new trial.
Before addressing the evidence, we must first determine the correct standard of review to be applied. While there are cases stating the standard in a variety of ways, not all of which are *571reconcilable; the correct standard is, in fact, well established. In Cohen v Hallmark Cards (45 NY2d 493 [1978]), the Court of Appeals explained the distinction between appellate review of the weight of the evidence and appellate review of the sufficiency of the evidence; in doing so, it instructed that as to a weight of the evidence review of a nonjury determination, the Appellate Division has the power to make new findings of fact: “In reviewing a judgment of Supreme Court, the Appellate Division has the power to determine whether a particular factual question was correctly resolved by the trier of facts. If the original fact determination was made by a jury, as in this case, and the Appellate Division concludes that the jury has made erroneous factual findings, the court is required to order a new trial, since it does not have the power to make new findings of fact in a jury case. The result is, of course, different in cases not involving the right to a jury trial, since then the Appellate Division does have the power to make new findings of fact. In either situation, the determination that a factual finding was against the preponderance of the evidence is itself a factual determination based on the reviewing court’s conclusion that the original trier of fact has incorrectly assessed the evidence” {id. at 498 [citations omitted and emphasis added]).
It has therefore become well settled that in reviewing a case tried without a jury, the Appellate Division’s “authority is as broad as that of the trial court” (Northern Westchester Professional Park Assoc. v Town of Bedford, 60 NY2d 492, 499 [1983]; see also 1 Newman, New York Appellate Practice § 4.03 [5], at 4-26). The Appellate Division “may render the judgment it finds warranted by the facts, taking into account in a close case the fact that the trial judge had the advantage of seeing the witnesses” (Northern Westchester Professional Park Assoc., 60 NY2d at 499 [internal quotation marks and citations omitted]).
Yet, defendant asserts that our review power is more limited here. It suggests that appellate review of nonjury determinations may be either de novo review, which it says is applicable where essentially legal issues were presented at trial, or weight of the evidence review, which it claims is appropriate where the determination under review was based on credibility, and which it characterizes as a more limited type of appellate review (citing Coliseum Towers Assoc. v County of Nassau, 2 AD3d 562 [2003]). It reasons that when the Court of Appeals remitted this matter for a “weight of the evidence” review, the Court intended to circumscribe this Court’s authority, and preclude a de novo review of the evidence. We reject this reasoning. To the extent some cases characterize weight of the evidence review as *572“limited” (see e.g. id.), we disagree. The Court of Appeals’ remittitur referred to a weight of the evidence review in order to distinguish that type of review from our prior determination, which was made on the law rather than on the facts.
Nor do we accept defendant’s suggestion that Thoreson v Penthouse Intl. (80 NY2d 490, 495 [1992]) dictates that our only task here is to decide whether the trial court’s determination was based on a fair interpretation of the evidence. The Thoreson decision concerned an award of punitive damages under Executive Law § 297 (9), and merely recited, without discussion, its agreement with the use of the “fair interpretation of the evidence” standard there. The questions raised in that case did not involve, and the Court neither discussed nor mentioned, the Appellate Division’s well established broad authority to make its own findings of fact, as recognized in Northern Westchester Professional Park Assoc. (60 NY2d at 499).
Moreover, the Thoreson decision specifies that the “fair interpretation” approach applies “especially when the findings of fact rest in large measure on considerations relating to the credibility of witnesses” (80 NY2d at 495 [internal quotation marks and citations omitted and emphasis added]). Limiting appellate review to the fair interpretation of the evidence approach may be appropriate where the findings rest predominantly on credibility determinations, because such determinations are entitled to substantial deference. However, it is not appropriate where the trial court’s findings rest largely on inferences drawn from established facts and verifiable assertions. In that case, there is no valid rationale for precluding the appellate court from finding facts, as indicated in Northern Westchester Professional Park Assoc. (60 NY2d at 499).
Here, although plaintiffs credibility was properly called into question by the trial court in some respects, when the entirety of the evidence is considered, it becomes apparent that the question of whether Mr. Green committed suicide is not logically dependent on findings regarding plaintiffs credibility. That is, our analysis does not turn on whether plaintiff was lying or telling the truth. Rather, this fact-finding determination is based predominately on inferences drawn from established facts such as empty pill vials and prescription dates, objectively verifiable assertions regarding the decedent’s conduct shortly before his death, and statements by witnesses whose credibility is not questioned. As to those aspects of plaintiffs testimony in which her credibility is arguably relevant to a finding, those assertions that are appropriately discounted or rejected based upon credibility problems do not have a significant impact on the question of whether Mr. Green committed suicide.
*573To conclude this preliminary discussion of the proper standard of review, we observe that since the Court of Appeals has already asserted that “the evidence [in this case] was strong enough to permit a finding of suicide, though not to require it” (12 NY3d at 347), there would be little point in further assessment if our task were limited to merely deciding whether the trial court’s determination was based on a fair interpretation of the evidence, rather than assessing de novo whether the weight of the evidence supports the determination.
In accordance with the foregoing, the standard of review we will apply here is the de novo weighing of the evidence set forth in Northern Westchester Professional Park Assoc., rather than the more limited approach referred to in Thoreson.
Before proceeding to weigh the evidence, we must also clarify defendant’s burden of proof. Plaintiffs burden of proof on her claim for the life insurance benefit is satisfied simply by proof of Mr. Green’s death, the existence of the life insurance policy, and plaintiffs status as the beneficiary of that policy (Schelberger v Eastern Sav. Bank, 93 AD2d 188, 192-193 [1983], affd 60 NY2d 506 [1983]). The claim that benefits are not payable because Mr. Green committed suicide constitutes an affirmative defense, which must be proved by defendant insurance company. Any affirmative defense—even one with no applicable presumption to overcome—places the burden of proof of that issue on its proponent (57 NY Jur 2d, Evidence and Witnesses § 164). Here, however, there is an additional burden on defendant. To establish the affirmative defense of suicide, an insurer must overcome a presumption that has been called “one of the strongest presumptions in the law” (Schelberger, 93 AD2d at 190). This burden has been said to require the insurer to establish suicide by “clearly establishing such facts as will exclude any reasonable hypothesis of accidental death” (id. at 192, quoting Vance, Insurance, at 571). Stated as the pattern jury instruction directs, the finding of suicide may be made only if the finder of fact is satisfied “that no conclusion other than suicide may reasonably be drawn” (PJI 4:57; 1:63.2). The Court of Appeals, in remanding this matter, approved the use of that instruction, although it also rephrased the instruction as one that tells jurors that suicide should not be found “unless the evidence shows suicide to be highly probable” (12 NY3d at 347). We therefore conclude that it is our obligation, just as it was the obligation of the trial court acting as factfinder, to apply the presumption against suicide in connection with defendant’s burden of proof on its affirmative defense, so as to find against suicide unless “no conclusion other than suicide may reasonably be drawn” (PJI 4:57) or *574“the evidence shows suicide to be highly probable” (12 NY3d at 347).
With these parameters in mind, we turn to the evidence.
As alluded to earlier in this discussion, we acknowledge, and to an extent agree with, the trial court’s view that in some respects plaintiffs trial testimony was not credible. Initially, we note that where the trial court questioned plaintiffs credibility based not on her courtroom demeanor—which this Court cannot observe and therefore could not rely on to question credibility—but on statements she is reported by others to have made shortly after her husband’s death, this Court is also capable of independently assessing plaintiffs credibility on that basis. Moreover, we would closely question plaintiffs credibility in any event, given her pecuniary interest in the matter.
The portion of plaintiffs testimony that warrants rejection on credibility grounds is the part in which she protested that her husband had not been depressed at the time of his death. This assertion was directly contradicted by her reported statements to police and to her sister-in-law on the day of Mr. Green’s death and the next day, that he had been depressed and that he must have overdosed on his medications. Moreover, the fact of Mr. Green’s depression was established by the unassailable testimony of his internist, Dr. Robert Bos, with whom he spoke the day before his death.
However, while plaintiffs unwillingness to acknowledge at trial her husband’s emotional difficulties may provide reason for rejecting her assertions on credibility grounds, it does not provide a basis to make affirmative findings of fact against her on the issue on which her adversary has the burden of proof, that is, that Mr. Green committed suicide. It is up to defendant to present evidence compelling that finding.
In an effort to establish that it has made such a showing, defendant characterizes as admissions plaintiffs expressions of fear, at the scene and shortly thereafter, that her husband must have died of a drug overdose. However, plaintiffs expressed fear or suspicion that her husband took an overdose of medication is not an admission of anything. Plaintiff’s statements may not logically be relied on either to establish the actual cause of his death or his intent at the time. Her expressions of fear or suspicion could satisfy defendant’s burden of proving that the feared possibility was a fact only if it were shown to be based on facts or events known to plaintiff and established at trial, that objectively support the conclusion that suicide, rather than accidental or unintentional death, was highly probable.
Another problem with the trial court’s finding that Mr. Green *575committed suicide is that the court improperly allowed, and then placed excessive reliance on, the testimony of defendant’s belatedly offered expert, forensic pathologist Dr. Michael Baden, when it found that “[t]he presence of suicidal thoughts in an individual is an important factor in determining whether the death of that individual was as a result of a suicide. The fact that an individual had been depressed in the immediate period before death is an important factor in determining whether the death of that individual was the result of suicide; and many suicides can be the results of acute reactive depressions which result from personal financial problems of a few days’ duration.” Dr. Baden had asserted that depression and suicidal thoughts are very important factors in making a diagnosis of suicide and that most suicides are not planned but are committed on the basis of opportunity.
Under the specific circumstances presented here, the ruling allowing defendant to present this testimony constituted an abuse of discretion.
CPLR 3101 (d) (1) provides that a party shall not be precluded from introducing an expert to testify at trial despite noncompliance with the statute’s notice requirement where the party has shown good cause for the belated application. The requirement of showing good cause has been considered satisfied where testimony offered by a witness at trial was entirely new and came as a surprise, such as in Simpson v Bellew (161 AD2d 693 [1990], lv denied 11 NY2d 808 [1991]), a personal injury action involving a pedestrian hit and killed by a van, in which a police officer called by the defendant testified for the first time at trial that the driver of the van told him that he had hit the pedestrian in the crosswalk, although no such admission had been noted in his police report. The surprise testimony not only was completely new, but it also was the type of information that would rationally be expected to be included in the police report, so the officer’s failure to report it before testifying at trial necessitated a new witness on the subject. In contrast, here, notwithstanding the defense’s characterizations, Dr. Bos’s testimony at trial contained nothing new.
While a trial court has wide discretion to allow a party to introduce expert testimony despite its failure to give the other side proper notice pursuant to CPLR 3101 (d) (see e.g. Putchlawski v Diaz, 192 AD2d 444, 445 [1993], lv denied 82 NY2d 654 [1993]), in the absence of prejudice (see St. Hilaire v White, 305 AD2d 209, 210 [2003]), here, the lack of prior notice of Dr. Baden’s testimony prejudiced plaintiff by leaving her unable to properly counter that testimony. Plaintiff should have been *576entitled, to rely on the absence of notice of a defense expert to conclude that she need not retain or consult her own expert beyond her husband’s treating physician, Dr. Bos.
One reason it is so troubling that plaintiff was prejudiced in this manner is that the situation defense counsel was attempting to solve with his sudden introduction of an expert witness was of his own making. It arose from defense counsel’s litigation decision to use Mr. Green’s treating internist, Dr. Robert Bos, on his direct case to establish that Mr. Green had been suicidal. Plaintiff did nothing to create the predicament in which the defense found itself. Since the burden was always on defendant to overcome the presumption and prove that Mr. Green committed suicide, and plaintiff had no burden on the issue, defendant cannot possibly point to plaintiffs not calling an expert to justify defendant’s initial decision not to call its own expert.
Moreover, the defense’s decision to prove through Dr. Bos that Mr. Green had been suicidal relied on a rather broad view of Dr. Bos’s deposition testimony. Dr. Bos testified at his deposition that Mr. Green said he had “suicidal thoughts,” but he further testified that Mr. Green immediately assured him that he did not want to kill himself, did not have plans to do so, and would never do such a thing.
Contrary to defense counsel’s characterization in the context of the mid-trial application to call Dr. Baden as a witness, Dr. Bos’s testimony at trial was not inconsistent with his deposition testimony. He testified at trial that Mr. Green “may not have cared about being alive at that point,” used words to the effect that he “[did not] feel life [was] worthwhile” and may have said he did not “see . . . the point of being alive.” Dr. Bos explained that it was based on such statements by Mr. Green that he made the notation “suicidal thoughts” in his records, but he explained how he differentiated between suicidal statements or thoughts and the state of actually being suicidal.
The purported contradictions defense counsel relied on in making the mid-trial application were not substantive contradictions and provided no actual support for the application. Defense counsel cited Dr. Bos’s failure to testify at trial that Mr. Green said he did not see the point of living, although he testified to that effect at his deposition. However, Dr. Bos’s trial testimony was virtually indistinguishable from his deposition testimony; to the extent he omitted mentioning at trial any particular statement attributed to Mr. Green at deposition, no direct contradiction was made out. Nor did Dr. Bos testify at trial, as defense counsel claimed, that “suicidal thoughts do[ ] not mean anything.” Rather, at both deposition and trial he discussed the *577statements Mr. Green made to describe how he then felt about his life.
Nor do defense counsel’s arguments on the present appeal support the claim that Dr. Bos changed his testimony, thereby making it necessary for the defense to call a new expert witness. The record does not support defendant’s contention that Dr. Bos tried to “distance himself’ from his earlier testimony characterizing Mr. Green as having suicidal thoughts, based on Mr. Green’s statement that he did not see the point of living.
The defense’s assertion that “Dr. Bos testified at trial that suicidal thoughts, without a plan to implement them, do not present a serious warning” is a distortion of the trial testimony. When we consider the testimony itself, as well as the manner in which it was elicited, it is clear that it cannot properly support a ruling allowing defense counsel to present a surprise expert witness. What occurred was that on redirect examination of Dr. Bos, defense counsel attempted to press its point that Mr. Green’s “suicidal thoughts” reflected that he was a suicide risk, by asking Dr. Bos a question more suited to an expert witness than to a fact witness. Specifically, defense counsel asked, “In somebody who is depressed, and . . . having suicidal thoughts, does that person present the same risk for suicide as a person who is depressed but is not having suicidal thoughts?” Dr. Bos replied that merely questioning the purpose of daily life does not, in itself, mean that a depressed person is going to take his own life. He added that it is “when they express to you a plan, and a concrete plan of really ending it all, then that would establish suicidality.”
While defense counsel clearly found this unexpected answer unsatisfactory, his unhappiness with Dr. Bos’s responses did not justify the court’s allowing him to bring in an expert in mid-trial. Dr. Bos’s answer did not contradict his earlier testimony. Rather, counsel asked him at trial a question he had not been asked before, and then did not like the answer. Moreover, since that portion of Dr. Bos’s testimony was elicited by defense counsel on a point not raised at deposition, on a subject more suited to an expert witness than to a fact witness, counsel should not have been permitted to rely on the unexpected answer to support his claim that he suddenly needed a new expert.
We also reject the suggestion of our dissenting colleague that the belated introduction of Dr. Baden was justified because Dr. Bos had purposely attempted to “weaken the implication that Mr. Green had committed suicide” by his testimony that merely questioning the purpose of life does not mean that a depressed person is going to take his own life. Dr. Bos was simply provid*578ing a fuller, more balanced and more nuanced answer to defense counsel’s question than the simple response counsel seemed to expect.
Finally, the minor discrepancies in Dr. Bos’s testimony as to who first told him over the telephone after Mr. Green’s death that Mr. Green had taken pills in a manner suggesting suicide fail to justify any relief. Indeed, while Dr. Bos expressed some uncertainty on this general subject, after his recollection was refreshed, he clarified that it was plaintiff who told him about the empty pill vial and the possibility of suicide.
In view of plaintiffs objection, the trial court should not have allowed defendant to present a new expert at that juncture. It was fundamentally unfair to allow the defense to bring in an expert witness in mid-trial when the sudden need for expert testimony was created by the defense’s strategic decision to attempt to establish through Mr. Green’s treating physician, a fact witness, a general truth about suicidal people, and that decision backfired.
The prejudice plaintiff experienced as a result of the surprise introduction of an expert in mid-trial was not eliminated by the offer of time for plaintiff to obtain a competing expert. In the midst of trial, attempting that task would entail an unacceptable diversion of counsel’s attention; as a practical matter, plaintiffs counsel could not undertake the task of locating a new expert to challenge Dr. Baden’s opinions and assertions as part of a rebuttal case. Counsel’s decision to decline the illusory offer of time was simply realistic, and should not be interpreted to mean that plaintiff was not prejudiced.
The ruling admitting Dr. Baden’s testimony is especially problematic because the trial court relied on it so heavily, particularly with regard to a questionable assertion by the expert that most suicides are not planned and are committed on the basis of opportunity. In fact, contrary to earlier research suggesting that many suicides are the result of impulsive decisions, recent research establishes that most suicides are not attempted impulsively, but involve a plan (see April R. Smith, Tracy K. Witte, Nadia E. Teale, Sarah L. King, Ted W Bender and Thomas E. Joiner, Revisiting Impulsivity in Suicide: Implications for Civil Liability of Third Parties, 26 Behav Sci & L 779 [Nov./Dec. 2008]). Nor was any explanation offered for permitting a forensic pathologist to testify as an expert on the psychology or state of mind of an individual who commits suicide. The resulting finding of suicide is particularly troubling, in the absence of evidence here tending to show any suicide plan on Mr. Green’s part, insofar as it was so heavily based on this *579surprise expert testimony that plaintiff was unable to effectively controvert.
Dr. Baden’s testimony must therefore be excluded in its entirety from the evidence to be considered in determining whether the verdict is supported by the weight of the evidence. Because that testimony provided by far the strongest evidentiary support for the finding that Mr. Green had committed suicide, and the remaining evidence consists largely of surmise, once this testimony is excluded from consideration, justification for the verdict is substantially undermined.
Even if we found that permitting Dr. Baden to testify did not constitute an abuse of discretion, we would nevertheless find that the trial court placed excessive reliance on Ms testimony, and in our present independent weighing of the evidence, we would, in any event, find that Dr. Baden’s testimony is entitled to little weight.
Another important component of the trial court’s finding of suicide was the inference the court drew from the empty pill vials that had contained Ambien and hydrocodone. The court calculated, based on the time it had taken Mr. Green to finish the prescription for 30 Ambien pills that he received on December 8, 2001 and refilled on February 6, 2002, that the amount of medication that would have been in the vial on the day he died was “inconsistent with an accident and only consistent with the fact that it was a deliberate suicidal act.” It further relied on the possibility that Mr. Green also took some of the 40 hydrocodone pills that had been prescribed for him in January after hernia surgery.
In our view, however, the conclusion that Mr. Green intentionally took an overdose of these two pills is based upon conjecture and is not sufficiently supported by the record. As to the painkiller hydrocodone, there is no basis for the conclusion that any of it remained in its vial by the date of his death, since it had been prescribed 27 days earlier and, if taken at anything like the prescribed rate of two every four hours, all 40 pills would have been taken well before that date. As to the Ambien, we simply cannot say how many pills remained in the Ambien vial by that date. Mr. Green’s earlier use of 30 Ambien pills during a previous 60-day period may be relevant, but cannot be relied upon by itself to establish as a fact his usage during the weeks preceding his death. Importantly, plaintiff said that Mr. Green took the Ambien regularly and that if he woke in the middle of the night, he took another pill or half a pill. She also admitted to having taken approximately five of the Ambien pills herself. This described usage could have left the vial empty or *580nearly empty on the date in question, without enough Ambien to cause death. But even if we do not credit plaintiffs description of how the Ambien was used, the mere fact that Mr. Green had been given 30 Ambien pills two weeks before his death creates, at best, a mere possibility that he had enough pills to overdose on them, not a circumstance that establishes a deliberate suicidal act.
Parenthetically, it seems perverse, to say the least, that a court would give greater credence to the contention that a drug addict who overdosed did so accidentally than to the suggestion that a nonaddict may have overdosed accidentally, as the trial court seemed to do in reliance on Schelberger v Eastern Sav. Bank (93 AD2d 188 [1983], supra).
Finally, the trial court acted improperly to the extent it determined that plaintiff was incredible based on the perceived inconsistency between her refusal to permit a toxicological exam or an autopsy of Mr. Green’s body on religious grounds and her arranging for Mr. Green’s remains to be cremated in accordance with his stated wishes, which the court asserted was in violation of those same religious tenets. It is presumptuous to term these two decisions inconsistent in support of a determination that plaintiff is not credible. Jews vary widely in their observance of Jewish law; while some attempt to strictly follow all 613 mitzvot in the Torah, others abide by far fewer. Each Jew makes an independent choice as to which of the mitzvot he or she will live by. There is nothing suspect in a Jewish person’s unwillingness to abide by particular tenets of Jewish law, and the decisions that person makes do not permit others to call into question that person’s character, sincerity or credibility. It is improper to find a Jewish person unworthy of belief simply because the person abides by some aspect of Jewish law but not another. This is what the trial judge did, and this is what Justice Andrias does as well. And, when the credibility determination based on the so-called inconsistency is examined in the sunlight and seen for what it is, a substantial chunk of the trial court’s findings falls away.
Moreover, the two decisions are not necessarily logically inconsistent. A Jew may express, while alive, a wish for his body to be cremated, without expressing any wish or preference concerning autopsies or toxicological exams. In such circumstances, after that individual’s death, the surviving relatives may feel bound by his expressed wish to be cremated, but, in the absence of any other direction about how his body should be treated, may feel authorized to make any remaining decisions based on their own views and observances.
*581The purported inconsistency therefore ought not serve as a basis for any sort of negative inference.
Nor is it appropriate to make a finding of suicide based on the conclusion that plaintiff sought to avoid the postmortem testing because she feared that an overdose would be discovered. The trial court reasoned that plaintiff did not permit the procedures because she “didn’t really want to find out” or was afraid of finding out that her husband did, in fact, commit suicide. However, this reasoning employs the same fallacy as defendant’s reliance on plaintiffs statements of fear that her husband had died of an overdose of his medications. Plaintiffs fear that her husband had committed suicide, and her purported desire to avoid having that fear confirmed, does not justify the inference that he committed suicide. It establishes neither the fact of an overdose nor that any such overdose was intentional rather than accidental.
We also reject defendant’s argument that plaintiffs refusal to consent to an autopsy or toxicological exam could not have been motivated by religious tenets, because if she had wanted to respect the family’s wishes, she would have consulted Mr. Green’s adult son or his sister, rather than his cousin. Nothing in the testimony reflects that Mr. Green was closer with his adult son or his sister than he was with his cousin, while there is evidence that Mr. Green and his cousin were close.
In addition, I find it objectionable that my colleague seems to implicitly draw a negative inference from plaintiffs failure to change her mind and grant permission for an autopsy and toxicology after the Deputy Medical Examiner advised her that it might be hard to collect on a life insurance claim in the absence of test results as to the cause of death. There is no reason why plaintiff should have reconsidered her decision based on the suggestion or advice of a medical examiner. In this context, my colleague also seems to imply that there was something untoward about the input of Mr. Green’s cousin, whom he refers to as “attorney Wolff,” in plaintiffs decision to refuse an autopsy and toxicology. It seems as though the term “attorney” is intended to raise the spectre of connivance and obfuscation. Any such implication is without any basis, however; the only evidence on the point shows Mr. Green to have been close to Mr. Wolff, which makes plaintiffs consultation with him nothing but appropriate.
Besides rejecting many of the underpinnings of the trial court’s finding of suicide, we observe that, notwithstanding the doubt cast on some of plaintiffs testimony, there is no reason to reject, and much evidentiary support for, plaintiffs testimony *582recounting her husband’s conduct on the morning of February 20, 2002, the day of his death. Indeed, the trial court accepted as fact plaintiff’s assertions that Mr. Green told her that he would be going to the gym that morning and that he had to make telephone calls, including a work-related conference call, that afternoon. Those assertions are confirmed by the fact, also found by the trial court, that when she found him on the bed that evening, he was dressed in gym clothes—-jeans, t-shirt and sweatshirt, with his sneakers on the floor next to the bed. Additionally, Mr. Green’s cousin, Richard Wolff, who was representing Mr. Green in litigation with his former employer, testified that he spoke with Mr. Green that morning, and that they scheduled a meeting for the following week. According to Mr. Wolff, Mr. Green was upbeat, positive and excited about the consulting business he had begun.
Furthermore, the testimony of Dr. Bos reflects that while Mr. Green was experiencing emotional difficulties, he was not overcome by them. Dr. Bos, upon hearing Mr. Green acknowledge that he was experiencing depression, anxiety and insomnia, directly inquired as to whether Mr. Green felt suicidal, and Mr. Green replied without qualification that “he would never do such a thing, he was not suicidal, he was just down.” We also observe that by going to the trouble of following up on his internist’s referral to a psychiatrist, with whom he left a voicemail message, Mr. Green demonstrated that he recognized, but refused to succumb to, his current state of depression.
The inference that Mr. Green’s death was unintended is further supported by additional facts as found by the trial court, including Mr. Green’s actions shortly before his death, such as contacting a psychiatrist, and the items found surrounding him at the time of his death, including a copy of the New York Times, his Palm Pilot and his portfolio. All these items, conversations and appointments point to a man engaged in life, not one determined to depart it.
As plaintiff reasonably suggested at trial, there are a variety of possible reasonable explanations for her husband’s death: It might have been caused by any number of sudden events such as a heart attack, an aneurysm, or an adverse reaction to medication. And if it was an overdose, it could just as easily have been accidental rather than intentional.
Weighing anew the entirety of the evidence, we find that the evidence tending to permit an inference of suicide is not sufficiently substantial to outweigh the strong presumption against suicide. We find suicide to be merely one possible cause of Mr. Green’s death but far from the only reasonable conclusion to *583reach. The presumption against suicide not being overcome, the weight of the evidence does not support the trial court’s finding, and a new trial is appropriate (Cohen v Hallmark Cards, 45 NY2d 493, 498-500 [1978], supra).
We recognize, of course, that only two members of this bench explicitly rule that the reversal we order should be based on the weight of the evidence; the concurring justice, declining to address the weight of the evidence, bases his determination that reversal is necessary on the improper introduction of an expert witness in mid-trial. However, it should not escape notice that the concurring justice has implicitly agreed with that portion of our plurality opinion which concludes that two important components of defendant’s case must be excluded when this court weighs the evidence. First, the conclusion that it was error to permit Dr. Baden’s testimony logically requires Dr. Baden’s testimony to be removed from the balance sheet. Second, by agreeing that plaintiffs expressions of her fears or beliefs with regard to how her husband died do not constitute affirmative proof of how he died, our colleague’s opinion precludes reliance on that testimony to support defendant’s claim of suicide. I submit that, even ignoring the other errors, simply removing those two components of defendant’s evidence from the balance sheet, particularly considering the centrality of Dr. Baden’s testimony, supports our factual finding that what remains is a puny quantum of evidence insufficient to overcome the ancient common-law presumption against suicide.
Accordingly, the judgment of the Supreme Court, New York County (Harold B. Beeler, J.), entered June 29, 2006, dismissing the complaint after a nonjury trial, reversed, on the facts, without costs, and the matter remanded for a new trial.
McGuire, J, concurs in a separate memorandum as follows: For the reasons stated by Justice Saxe, I agree that we should direct a new trial because Supreme Court erred in granting defendant’s mid-trial application to have Dr. Baden testify as an expert witness. In my view, the court abused its discretion in granting the application. In any event, I would substitute our discretion for that of Supreme Court and hold that Dr. Baden should not have been permitted to testify (see Brady v Ottaway Newspapers, 63 NY2d 1031 [1984]). As a new trial is necessary for this reason alone, there is no need to reach the issue of whether the verdict is against the weight of the evidence. But because there will be a new trial, I add that I also agree with Justice Saxe to the extent he concludes that the evidence concerning expressions by plaintiff of a fear or belief that her husband committed suicide are not affirmative evidence that he *584did commit suicide and that Supreme Court gave undue weight to that evidence.
I disagree with Justice Saxe’s view that I have “implicitly agreed with” him in two particular respects. My conclusion that Dr. Baden should not have been permitted to testify does not “logically require[ ] Dr. Baden’s testimony to be removed from the [weight-of-the-evidence] balance sheet.” First, evidence that should not have been admitted at trial is nonetheless evidence that was admitted at trial. Justice Saxe cites no authority for the proposition that when the weight of the evidence is assessed we must disregard evidence that was considered by the trier of fact on the ground that it should not have been admitted. In a criminal case, I think it plain that, for example, if we were to determine that an inculpatory statement of the defendant admitted at trial should have been suppressed, we would not appraise either the sufficiency or the weight of the evidence as if the statement had not been admitted. Nor can we assume there are no circumstances under which Dr. Baden (or another expert) might testify at the new trial. Second, the evidence relating to plaintiffs expressions of her fears or beliefs with regard to how her husband died may be admitted for impeachment purposes even though they are not substantive proof of how he died (see generally Barnes v City of New York, 44 AD3d 39, 47 [2007, Sullivan, J.], lv denied 10 NY3d 711 [2008]). To that extent, my opinion does not “preclude reliance on that [evidence] to support defendant’s claim of suicide.”