The defendant, Valerio DiGiambattista, was convicted of burning a dwelling house (G. L. c. 266, § 1). That conviction rested, in large measure, on DiGiambattista’s confession to the police during an unrecorded interrogation at a fire station. It is undisputed that, in an effort to obtain his confession, the interrogating officers resorted to trickery, falsely suggesting to DiGiambattista that his presence at the scene of the fire had been captured on videotape, while simultaneously expressing sympathy for his actions and opining that he needed counselling for his alcoholism. In his subsequent confession, DiGiambattista’s version of how and where he started the fire was completely contrary to the forensic evidence, and other details of his confession were ultimately shown to be impossible. On appeal, DiGiambattista contends that his motion to suppress the confession should have been allowed, and that, even with the introduction of the confession, the evidence against him was insufficient because the Commonwealth failed to present *425evidence corroborating that confession. See Commonwealth v. Forde, 392 Mass. 453, 458 (1984) (Forde). After remanding the case to the motion judge for further findings concerning the motion to suppress, the Appeals Court affirmed the conviction. Commonwealth v. DiGiambattista, 59 Mass. App. Ct. 190, 191, 199 (2003). We granted the defendant’s application for further appellate review, and we invited the filing of amicus briefs addressing whether we should expand the Forde corroboration rule and whether we should require electronic recording of custodial interrogations. For the following reasons, we conclude that the defendant’s confession should have been suppressed, and we therefore reverse the defendant’s conviction and remand the case for further proceedings. We also take this occasion to announce that, henceforth, the admission in evidence of any confession or statement of the defendant that is the product of an unrecorded custodial interrogation, or an unrecorded interrogation conducted at a place of detention, will entitle the defendant, on request, to a jury instruction concerning the need to evaluate that alleged statement or confession with particular caution.
1. Facts. The evidence, viewed in the light most favorable to the Commonwealth, was as follows. DiGiambattista, Nicole Miscioscia (his girlfriend and later fiancée), and Miscioscia’s children lived in a rented house at 109 Adams Street in Newton. The property was owned by Angelo Paolini, who had a construction company on the lot next door. During the final year of his tenancy, DiGiambattista withheld rental payments on account of Paolini’s failure to make much needed repairs to the premises.
On March 7, 1998, buoyed by the receipt of a tax refund, DiGiambattista and his family moved out of the 109 Adams Street property and relocated to an apartment in Chelsea. A few days prior to their departure, DiGiambattista installed a new lock on the front door, keeping one key for himself and giving the other one to his mother. (A few of his mother’s belongings remained on the premises, with the understanding that she would later retrieve them.) With that new lock in place, the only way to lock the door from the outside was by way of a key.
Shortly prior to midnight on Tuesday, March 10, 1998, a neighbor noticed smoke coming from the house at 109 Adams *426Street. When firefighters arrived at the scene, they found the front door locked, the other two doors boarded up, and the windows closed.2 Inspection of the scene and testing of samples revealed that the perpetrator had used gasoline as an accelerant and had started the fire in or near a closet underneath a stair landing. A second, but insignificant, fire had also been set in the kitchen sink, lighting a small amount of paper. Expert assessment was that the fire had been started sometime between 11:25 p.m. and 11:55 p.m.
The following evening, officers questioned DiGiambattista and Miscioscia concerning the fire. There were some inconsistencies in their versions as to DiGiambattista’s whereabouts on the night of the fire,3 and inconsistent versions as to who still had a key to the new front door lock. During initial questioning, DiGiambattista suggested the possibility that Paolini had set the fire, mentioning that Paolini may have used gasoline. (At that point in the interview, the officers had not said anything about an accelerant being used to start the fire, and did not yet have expert analysis identifying the accelerant as gasoline.) Further investigation uncovered a witness who claimed that he had seen a man resembling DiGiambattista enter the 109 Adams Street property at around 6 or 6:30 p.m. on the night of the fire.4
On April 10, 1998, one month after the fire, DiGiambattista voluntarily accompanied two officers (a State trooper and a Newton police officer) to a nearby fire station for further questioning. The trooper told DiGiambattista that he was free to leave, gave him Miranda warnings, and obtained his written waiver of rights. After an initial period of conversational, mild-mannered questioning, the officers changed tack and told Di*427Giambattista that he was their prime suspect, that his statements were inconsistent with those of other witnesses, and that they had a witness who placed him at the scene of the fire that night. DiGiambattista denied that he had been at the house that night and denied setting the fire. The officers inquired as to his willingness to take a lie detector test. After what the officers described as some hesitation and seeming reluctance, DiGiambattista agreed.5
At that point in the interview, another trooper came into the room carrying a thick folder and two videotapes. The folder was stuffed with blank paper and miscellaneous newspaper clippings having nothing to do with the case. One of the videotapes, marked “109 Adams Street,” was a recording made at the scene the night of the fire. The other, labeled “Paolini Construction Worker’s Comp Case,” was blank. This fictitious folder and videotape had been prepared in advance, with a plan that the other trooper would bring them into the interview room at a designated time. With the folder and videotapes conspicuously placed on the table next to the interrogating trooper, the trooper asked DiGiambattista: “If I told you that somebody at Paolini Construction was under surveillance by an insurance company for a workers’ comp fraud case, is there any reason you would show up on that videotape?” While confronting DiGiambattista with this ostensible evidence against him, the trooper simultaneously sought to “downplay the crime itself, and give [DiGiambattista] a way of saving face, to confess to this by downplaying it,” pointing out that “no one was hint” in the fire and that, in light of the deplorable condition of the premises, the trooper could “relate to” and “understand” his anger at the landlord and the desire to “do something like that.”6 DiGiambattista continued to deny that he had been at the scene.
The two officers left the room and were replaced by the *428trooper who had brought in the folder and tapes. That trooper repeated to DiGiambattista that they had a witness placing him at the scene, and that his own prior statement and the statements of witnesses contained inconsistencies as to the time he had left his new apartment that night (see note 3, supra). The trooper then expressed the view that DiGiambattista had not meant to hurt anyone, and that his lighting the fire was the product of stress, alcohol consumption, and understandable frustration with his living situation at 109 Adams Street. The trooper then gave DiGiambattista two proposals, that he had either done this “to hurt someone” or that he “had to be upset,” under “stress,” and that “when you add the booze in, you’re going to make mistakes.” During the trooper’s explanation, DiGiambattista began nodding, and then acknowledged that he “was stressed” and had been drinking. The trooper asked him if he had used matches or a lighter. DiGiambattista replied that he had used both. The trooper asked him how he had lit the fire. DiGiambattista said that he used gasoline. After one more question and answer (which the trooper could not recall at trial), the trooper summoned the original interrogators into the room to take a more detailed statement because they were the ones “who had all the background knowledge of the case.”7
DiGiambattista’s detailed statement to the officers recounted that he had traveled from Chelsea to 109 Adams Street using public transportation, stopping at a named hardware store in Watertown to buy a two and one-half gallon gasoline can. He also identified the gasoline station where he bought one dollar’s worth of gasoline. After a period of drinking beer, he proceeded to 109 Adams Street, let himself in the front door, and then *429poured gasoline in multiple locations in the house, “lighting the gas as [he] went along.” He claimed to have done so “in almost every room in the house.”
At the officers’ request, DiGiambattista also drew a diagram of where he had spread gasoline and lit fires. On the diagram of one floor, DiGiambattista wrote that he “[p] cured gas almost everywhere,” and on the diagram of the other floor, he indicated that he “[pjoured gas all over.” The diagram identified four definite locations (and a fifth possible location) where he had started fires on both floors. None of the locations identified in the diagram corresponded with the closet area where the experts determined the fire had been started. Despite the officer’s suggestion that DiGiambattista indicate the location of the kitchen sink in his diagram (an apparent attempt to jog his memory concerning the paper fire they had found in that location), DiGiambattista’s confession and diagram made no reference to starting a fire in the sink (or anywhere else in the kitchen). Similarly, whereas DiGiambattista unambiguously confessed to pouring gasoline liberally throughout both floors of the house, examination of the scene revealed gasoline only in a few specific locations, notably in the vicinity of the closet fire (an area not designated by DiGiambattista as a place he had spread gasoline).
In his confession, DiGiambattista said that he thought that he threw the gasoline can “in the back room,” but he was “not 100 percent sure.” He then “left out the front door and ran.” However, when the officers informed DiGiambattista that they had not found any gasoline can in the house, he changed his statement to say that he had taken the can with him on the bus as far as Harvard Square and had thrown it in a dumpster. The officers expressed incredulity that anyone would carry a gasoline can on a bus, at which point DiGiambattista changed his statement again, claiming that he had thrown the can away at a picnic area along the Charles River not far from the fire scene. DiGiambattista also told the officers that he had thrown the key in the river, but then later told them that he still had the key at home. (At the conclusion of the interrogation, the officers took DiGiambattista home, whereupon he retrieved the key and gave it to them.)
*430There was no electronic recording or taping of any portion of the interrogation or of the confession that it ultimately produced. One of the officers took down the confession in written form, which DiGiambattista reviewed, corrected, and signed. DiGiambattista also drew and signed the two-page diagram of both floors of the house described above. Finally, at the officers’ suggestion, DiGiambattista also composed a letter of apology for the fire, explaining that he had been “stressed out because of the conditions of the house” and “drunk,” expressing that he was “happy that nobody got hurt,” acknowledging that he “need[ed] help with [his] alcoholism problem,” and agreeing “to get help with [his] alcohol problem and any stress related problems.” The interrogation, lasting approximately two and one-half hours, was then concluded.
Subsequent investigation revealed that DiGiambattista could not have bought the gasoline can from the hardware store he had identified in his confession.8 Similarly, the gasoline station that he had identified had no record of anyone buying one dollar’s worth of gasoline that night. The officers later recovered a gasoline can from a back room at the house (on a different floor from the disposal location initially identified by DiGiambattista), but that can was a six gallon can (not the two and one-half gallon can described by DiGiambattista). There was no evidence of any gasoline can found in the picnic area where DiGiambattista had ultimately indicated he disposed of it.
2. Corroboration rule. At trial, the defendant moved for a required finding of not guilty, contending that the Commonwealth’s evidence failed to include sufficient corroboration of his confession, and that the Forde corroboration rule had therefore not been satisfied. The corroboration rule, as articulated in Commonwealth v. Forde, 392 Mass. 453 (1984), only requires corroboration that the underlying crime was in fact committed, thus preventing convictions against persons who have confessed to fictitious crimes. Id. at 458. Here, there was ample evidence that an arson had been committed, and, whatever the concerns about DiGiambattista’s confession (see *431note 15, infra), there was no suggestion that he was confessing to a crime that had not occurred. The trial judge correctly denied the motion for a required finding of not guilty.
On appeal, DiGiambattista asks that we expand the Forde rule to require corroboration that the defendant was the actual perpetrator of the crime, or a showing of circumstances that would make the confession itself reliable. See United States v. Lopez-Alvarez, 970 F.2d 583, 592 (9th Cir.), cert. denied, 506 U.S. 989 (1992). We agree with the Appeals Court that, even if we were to expand the rule in the manner DiGiambattista requests, any such expanded rule would have been satisfied in this case. Commonwealth v. DiGiambattista, 59 Mass. App. Ct. 190, 197 (2003). Beyond DiGiambattista’s confession, the Commonwealth contended at trial that DiGiambattista’s guilt was corroborated by the following: (1) DiGiambattista’s motive, namely, anger at his former landlord on account of the landlord’s failure to maintain the premises; (2) the witness who had seen someone resembling DiGiambattista on the premises at 6 or 6:30 p.m. on the night of the fire9; (3) evidence that the perpetrator had entered and left the premises by the front door, with DiGiambattista being one of only three people who had access to a key to that door10; and (4) DiGiambattista and others giving inconsistent statements about who had had the new front door key at various points over the course of their weekend move, and inconsistent versions as to whether and at what time DiGiambattista had left the apartment to run an errand on the night of the fire.
Some commentators have suggested that an expanded corroboration rule would help counteract the phenomenon of “false confessions.” See Ofshe, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 Denv. U. L. Rev. 979, *4321119 (1997). We question the efficacy of that approach. Police interrogations are not conducted at random, but often focus on persons who are already suspects, i.e., persons as to whom there is at least some basis for suspicion. To the extent that interrogation of an innocent suspect produces a “false confession,” ordinarily there will be at least some “corroboration” of that confession (i.e., the facts and circumstances that placed that person under suspicion in the first place). As such, requiring additional “corroboration” may do little to reduce the risk of a conviction predicated on a false confession. At this juncture, we decline to expand the corroboration rule, but instead focus our attention on other tools at our disposal to address the issue of alleged false confessions, namely, strict analysis of the circumstances of the interrogation as they affect the voluntariness of a defendant’s statement, and the suggestion that we take further steps to encourage or require electronic recording of interrogations.
3. Voluntariness of the confession. The principal issue raised with respect to the denial of the defendant’s motion to suppress is the voluntariness of his confession and, in particular, the calculated trickery that was used to obtain it. Twenty-five years ago, this court stated that “we expressly disapprove of the tactics of making deliberate and intentionally false statements to suspects in an effort to obtain a statement,” as “such tactics cast doubt” on both the validity of a suspect’s waiver of rights and the voluntariness of any subsequent confession. Commonwealth v. Jackson, 377 Mass. 319, 328 n.8 (1979). See Commonwealth v. Edwards, 420 Mass. 666, 671 (1995) (use of false statements to obtain suspect’s waiver is “disapproved of and may indicate that any subsequent waiver was made involuntarily”). See also Commonwealth v. Nero, 14 Mass. App. Ct. 714, 716 (1982) (“use of false information as a tactical device is strongly disapproved and casts instant doubt on whether a defendant’s statement is voluntary”). We have, however, repeatedly held that while the use of false statements during interrogation is a relevant factor on both waiver and voluntariness, such trickery does not necessarily compel suppression of the statement. See Commonwealth v. Scoggins, 439 Mass. 571, 576 (2003); Commonwealth v. Colby, 422 Mass. *433414, 416-417 (1996); Commonwealth v. Edwards, supra; Commonwealth v. Selby, 420 Mass. 656, 664 (1995); Commonwealth v. Forde, 392 Mass. 453, 455-456 (1984); Commonwealth v. Meehan, 377 Mass. 552, 563-564 (1979), cert. dismissed, 445 U.S. 39 (1980); Commonwealth v. Jackson, supra. See also Commonwealth v. Nero, supra at 716-718. Rather, the interrogator’s use of trickery is to be considered as part of the totality of the circumstances, the test that is used to determine the validity of a waiver and the voluntariness of any statement. Commonwealth v. Edwards, supra at 672. Commonwealth v. Forde, supra.
Close analysis of our case law on the subject of trickery suggests that where the use of a false statement is the only factor pointing in the direction of involuntariness, it will not ordinarily result in suppression, but that if the circumstances contain additional indicia suggesting involuntariness, suppression will be required. See Commonwealth v. Selby, supra at 662 n.1, 664-665 (despite false statement about retrieval of suspect’s handprint and fingerprint from scene, waiver and confession both voluntary where “all other relevant factors specific to the instant case indicate a voluntary waiver was made” and trickery was only factor suggesting involuntariness); Commonwealth v. Edwards, supra (confession admissible where nothing other than use of trickery would suggest involuntariness). See also Commonwealth v. Nero, supra at 718 (unintentional misstatement about strength of identification evidence against suspect did not render statement involuntary; where other evidence known to suspect was strong, single misstatement “had very little if any influence on his decision to talk to the police”). Contrast Commonwealth v. Meehan, supra at 562-564 & n.4 (police use of false statement concerning strength of identification testimony, combined with assurances of benefit from confessing, youth and emotional instability of suspect, and denial of right to use telephone, resulted in suppression); Commonwealth v. Jackson, supra at 322-324 (suppression required where police repeatedly failed to honor suspect’s invocation of right to remain silent, and used trickery to obtain subsequent waiver).
Here, there is no dispute that the trickery used was part of an *434orchestrated, prearranged plan amongst the officers — the videotape with its false label, and the artificially thick file, were both prepared in advance, with the intention that another trooper would bring them into the room at a particular point in the interrogation and that the interrogator would use them to convince DiGiambattista of the ostensible strength of the evidence against him. The Commonwealth contends that that trickery was the only factor suggesting involuntariness and that, as in the above cases, the trickery therefore does not compel suppression.11
While we adhere to the view that false statements about the evidence against the suspect do not automatically render the suspect’s confession involuntary, we note that ongoing research has identified such use of false statements as a significant factor that pressures suspects into waiving their rights and making a confession. See Comment, False Confessions and Fundamental Fairness: The Need for Electronic Recording of Custodial Interrogations, 6 B.U. Pub. Int. L.J. 719, 732-733 (1997); White, False Confessions and the Constitution: Safeguards Against Untrustworthy Confessions, 32 Harv. C.R.-C.L. L. Rev. 105, 145-149 (1997); Ofshe, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 Denv. U. L. Rev. 979, 1008-1013 (1997); Roppe, True Blue? Whether Police Should Be Allowed to Use Trickery and Deception to Extract Confessions, 31 San Diego L. Rev. 729 (1994); Kassin, The Psychology of Confession Evidence, 52 Am. Psychologist 221 (1997). This is particularly true where, as here, the false statement suggests a form of incriminating evidence that would be viewed as incontrovertible. If a suspect is told that he appears on a surveil*435lance tape, or that his fingerprints or DNA have been found, even an innocent person would perceive that he or she is in grave danger of wrongful prosecution and erroneous conviction.
We have recognized that false statements concerning ostensibly irrefutable evidence against a suspect are particularly troublesome when combined with suggestions of leniency in exchange for a confession. “[A] false statement concerning the strength of the Commonwealth’s case, coupled with an implied promise that the defendant will benefit if he makes a confession, may undermine ‘the defendant’s ability to make a free choice.’ ” Commonwealth v. Scoggins, 439 Mass. 571, 576 (2003), quoting Commonwealth v. Meehan, 377 Mass. 552, 563 (1979). “The specter of coercion arises in these circumstances from the possibility that an innocent defendant, confronted with apparently irrefutable (but false) evidence of his guilt, might rationally conclude that he was about to be convicted wrongfully and give a false confession in an effort to salvage the situation.” Commonwealth v. Scoggins, supra at 576-577, citing Ofshe, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 Denv. U. L. Rev. 979 (1997).
The motion judge found that, although there were various references during the interrogation to DiGiambattista’s need for “counseling,” the officers made no explicit promises of leniency in exchange for a confession to the crime. That finding, supported by the evidence, is one that we must accept.12 However, the absence of any explicit offer of leniency does not mean that there was no implied offer. Coercion may be readily applied by *436way of implied threats and promises, just as it is by express threats and promises.
References to DiGiambattista’s need for “counseling” implicitly suggested to him that “counseling” would be an appropriate avenue for him to pursue after making a confession, and that the interrogating officer would endorse such an approach. And, in addition to the references to “counseling,” the officers suggested to DiGiambattista that his commission of the crime was understandable, justifiable, excusable, and not that serious. Research suggests that such “minimization” of the crime by an interrogator implies leniency if the suspect will adopt that minimized version of the crime, and that leniency can thereby be implicitly offered even if it is not expressly stated as a quid pro quo for the confession. See S.M. Kassin and K. McNall, Police Interrogations and Confessions: Communicating Promises and Threats by Pragmatic Implication, 15 L. & Hum. Behav. 233 (1991). By the troopers’ own versions, “minimization” techniques were used repeatedly during this interrogation — the desire to bum the house was portrayed as understandable in light of the deplorable conditions that DiGiambattista and his family had had to endure during the years they lived in the house, DiGiambattista was under “stress,” he acted while under the influence of alcohol, and he had no intent to hurt anyone. This “minimization” of the crime, combined with references to DiGiambattista’s need for “counseling,” would at least imply to DiGiambattista that, if he adopted the troopers’ version of events, the crime would be viewed less seriously and could be resolved by mere “counseling.”13 The contention that this interrogation contained no express promise of leniency in exchange for a confession asks us to overlook evidence of “an implied promise that the defendant will benefit if he makes a confession” (emphasis added). Commonwealth v. Scoggins, 439 Mass. 571, 576 (2003).
Justices Spina’s dissent contends that research into the effects of “minimization” has been discredited and excluded from evidence, such that we should place no credence in it when as*437sessing the voluntariness of this confession.14 What has been excluded, by some courts, is expert testimony concerning the alleged relationship between various interrogation techniques and false confessions. See Vent v. State, 67 P.3d 661, 667-670 (Alaska Ct. App. 2003); State v. Cobb, 30 Kan. App. 2d 544, 552-553, 564-567 (2002); State v. Free, 351 N.J. Super. 203 (2002). Expert opinion whether a particular confession is “false” is problematic for a variety of reasons, most notably because it constitutes an opinion on the ultimate issue to be decided by a jury. However, where the opinion to be rendered has been carefully circumscribed to avoid that problem, other courts have allowed expert testimony on the subject of false confessions and interrogation techniques. See, e.g., United States v. Hall, 974 F. Supp. 1198, 1203-1206 (C.D. Ill. 1997), aff’d, 165 F.3d 1095 (7th Cir.), cert. denied, 527 U.S. 1029 (1999); People v. Page, 2 Cal. App. 4th 161, 179-189 (1991); Callis v. State, 684 N.E.2d 233, 239-240 (Ind. Ct. App. 1997).
More importantly, what we are considering today is not whether DiGiambattista’s confession was “false,” or whether he should have been allowed to present expert testimony on the subject of interrogation techniques. Rather, we are analyzing whether his confession was voluntary, and as part of that analysis, we must consider whether the interrogators communicated to DiGiambattista, expressly or implicitly, the impression that a confession would result in a lenient disposition. None of the cases cited by Justice Spina quarrels with the observation that the technique of “minimization” conveys that impression. Indeed, one of the reasons that courts have excluded expert testimony on this subject is that the impact of such interrogation techniques can be understood by jurors without expert *438assistance. See State v. Ritt, 599 N.W.2d 802, 812 (Minn. 1999) (expert testimony not needed where, from videotape of entire interrogation and confession, jury had “sufficient information” to evaluate defendant’s argument that interrogation was coercive); State v. Free, supra at 220-221 (“coercive factors” in interrogation techniques, including “minimization,” “are all matters that a jury would recognize as having a potential for causing a false confession”).
Even the principal proponents of these interrogation techniques do not dispute that suspects would interpret “minimization” as an offer of leniency, but rather dispute whether that implied offer of leniency would cause an innocent person to make a false confession. EE. Inbau, J.E. Reid, J.R Buckley, & B.C. Jayne, Criminal Interrogation and Confessions 419-422 (4th ed. 2004). Indeed, they recognize that most suspects need “tangible incentives” to confess, such as “a belief that the suspect will receive a lesser sentence if he fully cooperates, confesses, and expresses remorse for his crime,” that “[c]ommunicating these incentives in a legal manner” is necessary in order for the resulting confession to be admissible, and that interrogators therefore should not “explicitly state promises of leniency.” Id. They advocate “minimization” as a permissible method of creating the necessary “incentive” to confess, because it does not run afoul of the prohibition against explicit promises and would not (in their view) cause an innocent person to confess.
In short, common sense tells us that a person being asked by an interrogator to confess to a crime that is repeatedly described as understandable, justifiable, and not particularly serious would likely assume that giving the requested confession will result in lenient treatment. Scientific research has now confirmed the truth of that commonsense observation. Whether such an implied offer of leniency would in turn cause an innocent person to confess to that crime is a very different matter, and it is on that point that the research is still hotly debated and “not yet ready for ‘prime time.’ ” Post at 459 (Spina, J., dissenting), quoting Agar, The Admissibility of False Confession Expert Testimony, Army Law. 26, 42 (1999).
We do not suggest that an officer’s use of the standard inter*439rogation tactic of “minimization,” by itself, compels the conclusion that a confession is involuntary. Nor do we suggest that a mere “mention of counselling,” post at 456 (Spina, J., dissenting) by itself, would automatically constitute an improper offer of leniency undermining voluntariness. As always, such evidence must be considered as part of the totality of the circumstances. Here, those circumstances already included police resort to false statements concerning ostensibly irrefutable evidence of guilt, a tactic that “casts instant doubt” on the voluntariness of the subsequent confession. Commonwealth v. Nero, 14 Mass. App. 714, 716 (1982). When such trickery is then combined with multifaceted and repeated “minimization,” and then further combined with overt references to “counseling,” that “instant doubt” as to voluntariness is worsened, not dispelled.
On this motion, it was the Commonwealth’s burden to establish, beyond a reasonable doubt, that DiGiambattista’s confession was voluntary. Commonwealth v. Crawford, 429 Mass. 60, 65 (1999). Commonwealth v. Tavares, 385 Mass. 140, 151-152, cert. denied, 457 U.S. 1137 (1982). It is not up to a defendant to establish involuntariness. Here, the combination of trickery and implied promises, a combination that we have recognized as potentially coercive to the point of making innocent people confess to crimes, see Commonwealth v. Scoggins, 439 Mass. 571, 576 (2003), is such that the Commonwealth cannot meet its burden of proof on the issue of voluntariness.15 With all deference to the motion judge’s factual findings, we find error in the application of constitutional *440principles to those facts, and therefore reverse the order denying the defendant’s motion to suppress.
4. Electronic recording of interrogations. Although the above analysis suffices to resolve this particular appeal, the parties (and amici) have briefed the issue of requiring electronic recording of interrogations,16 and the present case serves as a useful illustration of the need better to preserve the details of an interrogation that results in a statement or confession by a suspect. As is all too often the case, the lack of any recording has resulted in the expenditure of significant judicial resources (by three courts), all in an attempt to reconstruct what transpired during several hours of interrogation conducted in 1998 and to perform an analysis of the constitutional ramifications of that incomplete reconstruction. Where, as here, it is the Commonwealth that bears the burden of proof, gaps in that reconstruction, and the inability to place the coercive features of the interrogation in their precise context, must result in suppression of the statement. We will never know whether, if able to hear (or even view) the entirety of the interrogation, the impact of the officers’ trickery and implied offers of leniency might have appeared in context sufficiently attenuated to permit the conclusion that DiGiambattista’s confession was nevertheless voluntary. “Given the fine line between proper and improper interrogation techniques, the ability to reproduce the exact state*441ments made during an interrogation is of the utmost benefit.” Commonwealth v. Scoggins, supra at 577 n.4 (where recording confirmed that officers had not used false evidence or made assurances of benefit from confessing, motion to suppress properly denied).17 We therefore take this occasion to revisit the issue of electronic recording of interrogations.
Eight years ago, this court announced that, although failure to record an interrogation would not result in automatic suppression of a defendant’s statement, the lack of a recording was itself a relevant factor to consider on the issues of voluntariness and waiver. Commonwealth v. Diaz, 422 Mass. 269, 273 (1996). See Commonwealth v. Burgess, 434 Mass. 307, 314 (2001); Commonwealth v. Larkin, 429 Mass. 426, 438-439 n.10 (1999). Put simply, a judge may reasonably conclude that when the party with the burden of proof beyond a reasonable doubt on the issues of voluntariness and waiver deliberately fails to utilize readily available means to preserve the best evidence of what transpired during the interrogation, it has not met that very high standard of proof. Just as failure to conduct a thorough investigation can itself comprise a basis for concluding that the Commonwealth has not met its burden to prove a defendant’s guilt beyond a reasonable doubt, see Commonwealth v. Bowden, 379 Mass. 472, 485-486 (1980), failure to preserve evidence of the interrogation in a thorough and reliable form can comprise a basis for concluding that voluntariness and a valid waiver have not been established beyond a reasonable doubt.18
While we have to date stopped short of requiring electronic recording of interrogations as a constitutional or common-law prerequisite to the admissibility of any resulting statements by the defendant, this court has repeatedly recognized the many benefits that flow from recording of interrogations. See Com*442monwealth v. Groome, 435 Mass. 201, 219 n.26 (2001); Commonwealth v. Burgess, supra at 314; Commonwealth v. Patterson, 432 Mass. 767, 778 n.14 (2000); Commonwealth v. Larkin, supra at 438 n.10; Commonwealth v. Fernandes, 427 Mass. 90, 98 (1998); Commonwealth v. Diaz, supra at 272-273; Commonwealth v. Fryar, 414 Mass. 732, 742 n.8 (1993), S.C., 425 Mass. 237, cert. denied, 522 U.S. 1033 (1997). Other jurisdictions, similarly reluctant to articulate a taping requirement as a matter of State constitutional law, have acknowledged that recording of interrogations would act as a deterrent to police misconduct, reduce the number and length of contested motions to suppress, allow for more accurate resolution of the issues raised in motions to suppress, and at trial on the merits, provide the fact finder a complete version of precisely what the defendant did (or did not) say in any statement or confession. See People v. Raibon, 843 P.2d 46, 49 (Colo. Ct. App. 1992); State v. James, 237 Conn. 390, 432, 434 (1996); State v. Kekona, 77 Haw. 403, 409 (1994); Stoker v. State, 692 N.E.2d 1386, 1390 (Ind. Ct. App. 1998); People v. Fike, 228 Mich. App. 178, 189-190 (1998) (Fitzgerald, J., concurring in part and dissenting in part); Williams v. State, 522 So. 2d 201, 208 (Miss. 1988); State v. Godsey, 60 S.W.3d 759, 772 (Tenn. 2001); State v. James, 858 P.2d 1012, 1018 (Utah Ct. App. 1993); State v. Kilmer, 190 W. Va. 617, 629 (1993). Commentators endorse the recording of interrogations, urging courts to adopt a requirement by way of court rule or constitutional law. See Westling, Something Is Rotten in the Interrogation Room: Let’s Try Video Oversight, 34 J. Marshall L. Rev. 537 (2001); Drizin, Let the Cameras Roll: Mandatory Videotaping of Interrogations Is the Solution to Illinois’ Problem of False Confessions, 32 Loy. U. Chi. L.J. 337 (2001); Comment, False Confessions and Fundamental Fairness: The Need for Electronic Recording of Custodial Interrogations, 6 B.U. Pub. Int. L.J. 719 (1997); White, False Confessions and the Constitution: Safeguards Against Untrustworthy Confessions, 32 Harv. C.R.-C.L. L. Rev. 105, 153-155 (1997); Leo, The Impact of Miranda Revisited, 86 J. Crim. L. & Criminology 621, 681-691 (1996); Kamisar, Foreword: Brewer v. Williams — A Hard Look at a Discomfiting Record, 66 Geo. L.J. 209 (1977); Unif. R. Crim. P. 243(b), *44310 U.L.A. 38 (Master ed. 2001); Model Code of PreArraignment Procedure § 130.4(3) (1975).
To date, only two State courts have imposed a requirement that interrogations be recorded. See Stephan v. State, 711 P.2d 1156, 1158 (Alaska 1985) (unexcused failure to record custodial interrogation violates due process clause of State Constitution); State v. Scales, 518 N.W.2d 587, 592 (Minn. 1994) (exercising court’s supervisory power to mandate suppression of unrecorded custodial interrogations).19 Meanwhile, three States and the District of Columbia have, by legislation, imposed a recording requirement for certain types of cases and interrogations. See 725 Ill. Comp. Stat. Ann. 5/103-2.1 (West 2003); Me. Rev. Stat. Ann. tit. 25, § 2803-B(l)(J), 2004 Me. Legis. Serv. 780 (West 2004); Tex. Code Crim. Proc. Ann. art. 38.22, § 3 (West 1999); D.C. Code Ann. § 5-133.20 (2003). Despite initial reluctance on the part of law enforcement personnel, actual experience with recording of interrogations has confirmed that the benefits expected from the procedure have indeed materialized, and most of those benefits ultimately inure to the prosecution, not to the defendant. Sullivan, Police Experiences with Recording Custodial Interrogations, Nw. U. Sch. Law, Center on Wrongful Convictions, Special Report (2004); Geller, Videotaping Interrogations and Confessions, National Institute of Justice, U.S. Dep’t of Justice, Research in Brief (Mar. 1993); Drizin, supra at 393-400.
The principal objection to recording of interrogations springs from the fear that suspects will refuse to talk at all, or will decline to make a fuU confession, if they know they are being recorded. Based on experience to date in other jurisdictions, those fears appear exaggerated.20 Moreover, what is posed as an objection to recording of interrogations is itself inherently *444contrary to our requirement of a knowing and voluntary waiver of the right to remain silent. We recognize that interrogating officers may prefer to lull suspects into the mistaken belief that they are having a confidential chat with a sympathetic listener, and thus do not want an ongoing audiotape or videotape recording that will serve to remind a suspect of the fact “that anything he says can be used against him in a court of law.” Miranda v. Arizona, 384 U.S. 436, 479 (1966). A technique, like recording, that reinforces a suspect’s understanding and appreciation of that portion of the Miranda warnings is not to be eschewed because it would have that desirable reinforcing effect.21
Those opposing the imposition of any requirement that interrogations be recorded contend that, whatever the benefits of recording, it is beyond this court’s power to regulate the activities of law enforcement, and that attempts to do so would violate the separation of powers. See art. 30 of the Declaration of Rights of the Massachusetts Constitution. The issue, however, is not what we “require” of law enforcement, but how and on what conditions evidence will be admitted in our courts. We retain as part of our superintendence power the authority to *445regulate the presentation of evidence in court proceedings. The question before us is whether and how we should exercise that power with respect to the introduction of evidence concerning interrogations.
Proponents of electronic recording of interrogations ask that we, like Minnesota, exercise our superintendence power to impose a bright-line rule refusing to admit in evidence statements and confessions obtained by way of unrecorded custodial interrogation. See State v. Scales, supra. Although appealing in its superficial simplicity (and unquestionably an effective method of convincing law enforcement officials to adopt recording as a standard practice), we still decline to impose such a rule. Among other problems, adoption of a rule excluding evidence of unrecorded interrogations necessitates precise identification of what interrogations will be subject to that rule — does it cover only custodial interrogations, or should it also cover any noncustodial interrogation conducted in particular locations (e.g., at police stations)? If the requirement were to be premised on the custodial (as opposed to noncustodial) nature of the interrogation, what do we do with interrogations that start out as noncustodial but arguably become custodial at some later (and often disputed) point during questioning? A rule of exclusion would also have to allow for justifiable failures to record — e.g., equipment malfunction, or the suspect’s refusal to allow recording (or insistence that the tape recorder be turned off at a particular point during the interrogation). See G. L. c. 272, § 99. With regard to a suspect who is willing to speak to the interrogator but initially unwilling to be recorded, would we need to impose some requirement that the interrogator make a good faith effort to convince the suspect to agree to recording, lest that ostensible “justification” for not recording too easily become the exception that swallows the rule? Notwithstanding predominantly positive experiences in those jurisdictions that have imposed recording requirements as a prerequisite to admissibility, we are hesitant to formulate a rigid rule of exclusion, and all its corollary exceptions and modifications (each of which would potentially spark new disputes in motions to suppress).
We are not, however, satisfied with preservation of the status quo, which amounts only to repeated pronouncements from the *446court about the potential benefits of recording interrogations. Just as we have advised judges of the significance they may attach to the lack of a recording when deciding motions to suppress, we believe it is appropriate to provide juries with that same advice.
There is no dispute that the evidence of a defendant’s alleged statement or confession is one of the most significant pieces of evidence in any criminal trial. When there is a complete recording of the entire interrogation that produced such a statement or confession, the fact finder can evaluate its precise contents and any alleged coercive influences that may have produced it. Where, however, the fact finder is presented only with officers’ testimony as to what they recall as the significant aspects of the interrogation and the defendant’s statement (and perhaps a defendant’s equally selective and biased account of the same interrogation and statement), the fact finder has a woefully incomplete and inherently unreliable version of what everyone recognizes as critical evidence in the case.22 Even assuming the most conscientious and good faith efforts of an interrogating officer, and even aided by a contemporaneous written statement or summary report, the officer can at best reconstruct only a portion of what was said over the course of an interrogation conducted months and oftentimes years prior to the time the officer testifies. Those contemporaneous written reports and statements can only reflect what the officer at the time perceived to be of significance to the case, which may not include issues that emerge only later during the investigation or at trial. The failure better to preserve this critical evidence in the first place, a failure that is often attributable to the strategic decision of the interrogating officer, merits the fact finder’s express consideration.
Where, as here, there are grounds for questioning the reliability of certain types of evidence that the jury might misconstrue as particularly rehable, specific instruction to the jury may be appropriate. See, e.g., Commonwealth v. Ciampa, *447406 Mass. 257, 263-264, 266 (1989) (cautionary instruction required when jury hears that plea or immunity agreement requires witness to give “truthful” testimony). We recognize the exceptionally potent quality of a defendant’s statement or confession, and that potency can only be magnified when the evidence of that statement or confession is presented by one or more of the interrogating officers. Where, however, those interrogating officers have chosen not to preserve an accurate and complete recording of the interrogation, that fact alone justifies skepticism of the officers’ version of events, above and beyond the customary bases for impeachment of such testimony. We believe that a defendant whose interrogation has not been reliably preserved by means of a complete electronic recording should be entitled, on request, to a cautionary instruction concerning the use of such evidence.23’24
Thus, when the prosecution introduces evidence of a defendant’s confession or statement that is the product of a custodial interrogation or an interrogation conducted at a place of detention (e.g., a police station), and there is not at least an audiotape recording of the complete interrogation, the defendant is entitled (on request) to a jury instruction advising that the State’s highest court has expressed a preference that such interrogations be recorded whenever practicable, and cautioning the jury that, because of the absence of any recording of the inter*448rogation in the case before them, they should weigh evidence of the defendant’s alleged statement with great caution and care. Where voluntariness is a live issue and the humane practice instruction is given, the jury should also be advised that the absence of a recording permits (but does not compel) them to conclude that the Commonwealth has failed to prove voluntariness beyond a reasonable doubt. See Commonwealth v. Cryer, 426 Mass. 562, 571 (1998), and cases cited (jurors must disregard defendant’s statement if voluntariness not established beyond a reasonable doubt).
Nothing in this instruction alters the overarching requirement that the voluntariness of a defendant’s statement be determined on the totality of the circumstances. Commonwealth v. Selby, 420 Mass. 656, 662-663 (1995), and cases cited. To the contrary, the instruction aptly focuses the jury’s attention on the fact that the Commonwealth has failed to present them with evidence of the “totality” of the circumstances, but has instead presented them with (at best) an abbreviated summary of those circumstances and the interrogating officers’ recollections of the highlights of those circumstances. Jurors should use great caution when trying to assess the “totality of the circumstances” when they have before them only a highly selective sliver of those circumstances, and they may properly decide that, in the absence of that “totality,” they cannot conclude that the defendant’s statement was voluntary.25
Where we now mandate a jury instruction, not a rule of exclusion, we think that the instruction is appropriate for any custodial interrogation, or interrogation conducted in a place of detention, without regard to the alleged reasons for not recording that interrogation.26 It is of course permissible for the prosecution to address any reasons or justifications that would *449explain why no recording was made, leaving it to the jury to assess what weight they should give to the lack of a recording. The mere presence of such reasons or justifications, however, does not obviate the need for the cautionary instruction.
Nor, in light of our repeatedly announced preference that interrogations be recorded and the widespread recognition of the benefits of recording interrogations, do we see the need to postpone the implementation of our decision. Law enforcement officials who have chosen not to avail themselves of the opportunity to preserve this critical evidence cannot complain that we have decided to highlight to juries the potential significance of that choice.
Despite our view that recording all interrogations would improve the efficiency, accuracy, and fairness of criminal proceedings, we still decline at this time to make recording of the interrogation a prerequisite to the admissibility of a defendant’s statement. However, where the utilization of recording is left to the unfettered discretion of law enforcement (as it is at present), and an officer has chosen not to record a particular interrogation, we think that it is only fair to point out to the jury that the party with the burden of proof has, for whatever reason, decided not to preserve evidence of that interrogation in a more reliable form, and to tell them that they may consider that fact as part of their assessment of the less rehable form of evidence that the Commonwealth has opted to present.
5. Conclusion. The judgment of conviction is reversed, the verdict is set aside, and the case is remanded to the Superior Court for further proceedings consistent with this opinion.
So ordered.
There was evidence that a rear window at ground level may have been open, but one of the officers testified that there was plastic covering that window.
Specifically, Miscioscia told the officers that DiGiambattista never left the Chelsea apartment at all the night of the fire, whereas DiGiambattista himself acknowledged that he had left on a brief errand at around 10:30 p.m., and a visitor to the apartment reported his having left on his errand much earlier in the evening. Video surveillance at an automated teller machine in Chelsea ultimately confirmed that DiGiambattista had made a withdrawal there at 7:50 p.m. on the night of the fire.
The witness had not seen the face of the person entering the house and did not make a positive identification of DiGiambattista.
No polygraph test was administered, and the officers had no intention of administering one. They posed the inquiry in order to gouge DiGiambattista’s reaction to the suggestion.
At the hearing on the motion to the suppress, the officers denied that they had promised to get counselling for DiGiambattista if he confessed, but the principal interrogator acknowledged telling DiGiambattista that he “probably needs some type of counselling.”
At the hearing on the motion to suppress and at trial, DiGiambattista testified that other pressures were brought to bear to obtain his confession, namely, that the troopers threatened to arrest both him and Miscioscia, warned that a high bail would keep him in custody over the weekend (with an even higher bail likely to be imposed at arraignment on Monday), and predicted that State officials would take their children away. He also testified that the officers told him that, if he confessed, the case could be resolved with his participation in counselling. He also claimed that he asked for an attorney, but was told that the involvement of any attorney would be the end of the “deal.” The motion judge’s findings (rendered on remand four and one-half years after the evidentiary hearing on the motion) did not credit DiGiambattista’s testimony on any of these points.
That store had closed at 8 p.m. on the night of the fire (at which time DiGiambattista was undisputably still in Chelsea), and its records indicated that no one had bought a gasoline can that day.
It was undisputed that DiGiambattista was at a bank in Chelsea at 7:50 p.m., but he could have made the trip from Newton to Chelsea in that time and then returned to Newton later that night. However, DiGiambattista’s confession makes no mention of being at the premises that evening at any time prior to lighting the fire.
This theory presupposed that DiGiambattista, ostensibly drunk and fleeing a house he had just set on fire, had stopped to relock the front door behind him. DiGiambattista’s confession made no reference to locking the door as he “ran” from the house.
The Commonwealth also suggests that the trickery was less potent because the trooper did not affirmatively state that DiGiambattista had been identified on the videotape, but only asked a question (“If I told you that . . . ?”). Without regard to the precise grammatical construct of the trooper’s reference to the videotape, the question inherently conveyed to DiGiambattista that the police had a solid, videotaped identification of him at the scene on the night of the fire. With the deliberately planted “prop” of a videotape carrying a fake label that corresponded to the premise of the trooper’s “question,” the obvious purpose was to make DiGiambattista believe that his likeness was indeed on the videotape before him.
The judge’s findings do not adopt the sequence of events that Justice Spina’s dissent assumes, namely, that no references to “counseling” were made until after DiGiambattista confessed. Post at 456 (Spina, J., dissenting). The principal interrogator’s testimony made no reference to such a sequence, and his testimony was given in response to a series of questions addressing what was said to DiGiambattista that would potentially convince him to confess. While another trooper testified that a discussion of “counseling” occurred after DiGiambattista’s confession, that trooper had not been in the room during the principal interrogator’s questioning of DiGiambattista, and thus would not have heard any earlier references to “counseling.” The judge discredited DiGiambattista’s claim that counselling was expressly offered in exchange for his confession, but did not discredit the testimony that the subject of “counseling” came up during the interrogation.
Not surprisingly, DiGiambattista’s confession and letter of apology adopted every single one of the minimizing themes suggested by the officers, acknowledged his need for “help,” and promised to obtain that “help with [his] alcohol problem and any stress related problems.”
Justice Spina also suggests that the author of these studies has disavowed his own research, quoting his statement that the experiments yielded only “mixed support for the hypothesis that minimization may pragmatically imply an offer of leniency.” Post at 458 (Spina, J., dissenting). The author was explaining that an initial series of general experiments had yielded that “mixed support,” whereupon he designed and conducted further experiments to focus on that precise question. Those subsequent experiments confirmed not only that “minimization” conveys an offer of leniency, but that it does so just as effectively as an explicit offer of leniency. Kassin, Police Interrogations and Confessions: Communicating Promises and Threats by Pragmatic Implication, L. & Hum. Behav. 233, 239-241 (1991).
Although voluntariness is assessed based on the totality of the circumstances surrounding the interrogation, and not on the reliability of the confession itself, it is nevertheless troublesome that this particular confession has been shown to be inaccurate in so many important respects. Significant discrepancy between the known facts of the crime and the details of a confession is recognized as an indicator that a confession may be false. See Ofshe, The Decision to Confess Falsely: Rational Choice and Irrational Action, 74 Denv. U. L. Rev. 979, 1118-1120 (1997). Indeed, as one technique to guard against use of false confessions, it is recommended that interrogators get the confessor to give his or her own narrative version of the details of the crime, so that they may be compared with details uncovered in the investigation that would not be known to anyone other than the perpetrator. Id. Here, the officers followed that recommendation, and obtained DiGiambattista’s own description of how and where he ostensibly poured gasoline and lit the fire. However, that description was totally *440at odds with the forensic evidence, and his detailed accounts of where and how he got the gasoline can and the gasoline were equally impossible. The motion judge recognized that there were discrepancies between DiGiambattista’s version and the known facts, but chose to explain them by reference to DiGiambattista’s intoxication at the time of the fire. Of course, the only evidence that DiGiambattista was intoxicated that night comes from the confession itself, and the troopers had repeatedly suggested to DiGiambattista that “booze” would make his crime more understandable. Moreover, while intoxication might indeed impair a perpetrator’s memory as to precise locations and sequence, DiGiambattista’s version is not merely slightly off the mark, or a bit confused or incomplete, but is hopelessly contrary to every known fact as to how this fire was set and lighted. We are not in a position to determine whether this particular confession was “false.” (Indeed, the jury, having heard all the evidence, were convinced that the confession was “true.”) However, at least one recognized indicator of a possible false confession is readily apparent in this case.
We acknowledge amicus briefs filed by the Attorney General, the Secretary of Public Safety, the Massachusetts District Attorneys, and the Boston police department; Suffolk Lawyers for Justice, Inc., and the New England Innocence Project; and Committee for Public Counsel Services.
Indeed, it is ironic that, in our two leading cases upholding the admissibility of a confession despite police resort to trickery, significant portions of the interrogation or confession were recorded. See Commonwealth v. Edwards, 420 Mass. 666, 669 (1995); Commonwealth v. Selby, 420 Mass. 656, 658-659 (1995).
Notwithstanding our articulation of that view, and notwithstanding that DiGiambattista raised the issue during the evidentiary hearing, the motion judge’s findings and conclusions did not address or mention the lack of a recording.
In the exercise of its supervisory authority, the Supreme Court of New Jersey has recently established a committee to investigate whether to “encourage” electronic recording of interrogations by imposing “a presumption against admissibility of a non-recorded statement.” State v. Cook, 179 N.J. 533, 562 (2004). One Justice of that court expressed the view that a recording requirement should be implemented immediately. Id. at 569 (Long, J., dissenting).
For reasons that are unclear, Justice Spina’s dissent accords greater credence to the speculative fears of investigators who do not normally record interrogations, post at 459-460 (Spina, J., dissenting), than to studies reflecting *444interviews with hundreds of law enforcement agencies who have years of experience recording interrogations as a regular practice. See Sullivan, Police Experiences with Recording Custodial Interrogations, Nw. U. Sch. Law, Center on Wrongful Convictions, Special Report (2004); Geller, Videotaping Interrogations and Confessions, National Institute of Justice, U.S. Dep’t of Justice, Research in Brief (Mar. 1993). Both of those studies reflect that, notwithstanding initial apprehension and skepticism, law enforcement agencies overwhelmingly endorse the practice of recording interrogations once they have gained experience with it.
The issue of financial cost, although raised in a footnote in an amicus brief, has not been identified as a significant obstacle to recording interrogations. To the extent that there are some police departments or law enforcement agencies that do not already have recording equipment, the cost of the equipment is minimal, and that cost is dwarfed by comparison to the costs of having officers spend countless hours testifying at hearings and trials in an attempt to reconstruct the details of unrecorded interrogations. See Commonwealth v. Patterson, 432 Mass. 767, 778 n.14 (2000) (noting that recording of interrogation of murder suspect would have avoided conflict of interest with attorney witness and thereby avoided necessity for retrial). Again, law enforcement agencies with significant experience recording interrogations confirm that the benefits of the practice far outweigh the costs. See note 20, supra.
Proponents of recording correctly point out that other critical evidence is ordinarily preserved with far greater care — e.g., crime scenes are photographed or even (as here) videotaped, not just described from the witness stand by the responding officers.
We do not mean to suggest that the prosecutor must introduce the entirety of a recorded interrogation in evidence in order to avoid the cautionary instruction. Indeed, there will often be times that neither side wants the finder of fact to hear (or the public record to reflect) the entirety of the recording, and ordinary evidentiary rules may operate to exclude portions of a recording. As long as the entire recording is available to the defendant, issues as to the reliability and completeness of any testimony concerning the interrogation can be adequately addressed, and there is no need for any cautionary instruction. Rather, the instruction is predicated on the failure to preserve the evidence in the first place, not by any failure to introduce it.
We note that a similar approach was adopted by the High Court of Australia, which required that a cautionary instruction be given to the jury “whenever police evidence of a confessional statement allegedly made by an accused while in police custody is disputed and its making is not reliably corroborated.” McKinney v. The Queen, 171 C.L.R. 468, 475 (1991). Legislation requiring the recording of certain interrogations was later enacted. See Westling, Videotaping Police Interrogations: Lessons from Australia, 25 Am. J. Crim. L. 493, 529-534 & n.236 (1998).
Contrary to the fears expressed in Justice Greaney’s dissent, this is not a “dynamite” charge that will cause automatic rejection of confession evidence. Post at 451 (Greaney, J., concurring in part and dissenting in part). Jurors receive cautionary instructions on various subjects, some of them very strongly worded. For example, we have no reason to believe that the instruction required by Commonwealth v. Ciampa, 406 Mass. 257, 263-264, 266 (1989), has resulted in the wholesale rejection of testimony from cooperating witnesses.
For now, this instruction need be given only in cases involving interrogation of a defendant in custody or at a place of detention. While recordings of *449interviews in other settings may also be desirable, we recognize that recording is not practicable in all circumstances. However, if an interrogation is conducted at a police station, or if the interrogation is custodial, it is reasonable to expect that arrangements for recording can be made.