O'CONNOR, J.

In this case, we reexamine the admissibility of polygraphic evidence in criminal trials in this Commonwealth. Persuaded both by the failure of the basic theory of polygraphy to have gained general acceptance among physiological and psychological authorities, and by the nearly unanimous rejection of such evidence by courts throughout the United States (at least in the absence of stipulation), we conclude that polygraphic evidence is inadmissible in criminal trials in this Commonwealth either as substantive proof of guilt or innocence or as corroboration or impeachment of testimony.

The defendant Mendes is charged with rape of a child, indecent assault and battery on a child under sixteen, and rape of a child by force. The defendant Rosenberg is charged with rape of a child (two indictments), incest, and indecent assault and battery on a child under fourteen (three indictments). The defendants moved for court-ordered polygraphic examinations. Also, the defendant Mendes filed a motion seeking admission in evidence of the results of his previously court-ordered polygraph test. The motions were heard together by a judge of the Superior Court at an evidentiary hearing that consumed four days.

The judge issued a thorough memorandum in which he discussed the evidence at length, including numerous written studies, and set forth his findings and conclusions. He concluded as follows: "[T]he polygraph is sufficiently reliable to warrant its continued limited admissibility provided that any court-ordered examination is subject to testing by the traditional tools of the adversary system; namely discovery, cross-examination, and rebuttal. Discovery of a defendant's previous polygraph history, his knowledge of countermeasures, and his criminal, social and psychiatric history might provide evidence for meaningful cross examination and a basis for expert rebuttal and surrebuttal testimony. However, in order to give fair and appropriate weight to the results of an individual court-ordered test, the cross-examination of the defendant and the polygraph operator on these issues could be extensive. An expert challenging the test results in rebuttal and another supporting the test results in surrebuttal may cover the same ground and, in fact, parallel the four day hearing conducted by this court. In essence, this Court is concluding that the polygraph is valid, but that the necessary evaluative time and resources may be so substantial, that an appellate authority may, on policy grounds, decide that it is not worth the price." (Emphasis in original.) The judge concluded as follows: "With full discovery of the defendant's polygraph history and a broadened line of inquiry at trial concerning this history, the Court finds that the polygraph, although it has not gained general acceptance in the scientific community, is sufficiently reliable for its continued use under the procedures authorized by Commonwealth v. Vitello [, 376 Mass. 426 (1978)]. . . . Broadening the scope of the in-court inquiry concerning the weight to be given the court-ordered test will place significant burdens on the system which should be addressed by an appellate authority as a matter of policy."

The judge allowed each defendant's motion for a court-ordered polygraph examination, subject to conditions, among which are the requirements that the defendant file with the clerk "the results of any prior polygraphy test or tests he has taken along with an affidavit detailing his previous experience with the polygraph," and that those results and affidavit "be made available to the court-ordered polygrapher."

At the Commonwealth's request, the judge reported the following questions of law to the Appeals Court:

"1. Should the polygraph continue to be admissible for the limited purpose of corroborating or impeaching a defendant's trial testimony in view of the validity research and expert opinion since the decision in Commonwealth v. A Juvenile, 365 Mass. 421 (1974)?

"2. If the answer to one above is yes: `In view of the research and expert opinion, does the taking of a private polygraph examination invalidate a later court-ordered test?'

"3. If the answer to two above is no: `Can the trial judge order disclosure of the results of a privately retained preliminary test as well as other information concerning previous polygraph knowledge and experience possessed by the defendant?'

"4. If the answer to three above is yes: `Is this information admissible at trial on the issue of the weight to be given to the court sanctioned test?'

"5. If the answer to one above is yes: `Do special circumstances such as the nature of the offense charged, the criminal and psychiatric history of the defendant, or the use of alcohol or drugs at the time of the events invalidate the test?'"

We granted the Commonwealth's application for direct appellate review. For the reasons stated below, we answer the first question "no; evidence that a defendant has taken a polygraphic examination, or testimony as to the results of such an examination is inadmissible at a criminal trial." Thus, we need not answer the remaining questions. We vacate the order allowing the defendants' motions for court-ordered polygraph tests.

In Commonwealth v. Fatalo, 346 Mass. 266 (1963), we first addressed the question whether the results of a polygraph test should be admissible in evidence at a criminal trial. In answering that question in the negative, we adopted the rule articulated in the landmark case of Frye v. United States, 293 F. 1013, 1014 (D.C. Cir. 1923), that, "while courts will go a long way in admitting expert testimony deduced from a well-recognized scientific principle or discovery, the thing from which the deduction is made must be sufficiently established to have gained general acceptance in the particular field in which it belongs." In rejecting an early predecessor of polygraphy, the Frye court stated: "We think the systolic blood pressure deception test has not yet gained such standing and scientific recognition among physiological and psychological authorities as would justify the courts in admitting expert testimony deduced from the discovery, development, and experiments thus far made." Id. In Fatalo, supra at 270, we, too, concluded that the polygraph test had not yet been accorded general scientific recognition, and that, therefore, the trial judge had properly excluded such evidence.

"The requirement, as in the Frye and Fatalo cases of general acceptance in the scientific community assures that those most qualified to assess the general validity of a scientific method will have the determinative voice. See United States v. Addison, 498 F.2d 741, 743-744 (D.C. Cir. 1974)." Commonwealth v. Lykus, 367 Mass. 191, 202 (1975). Those most qualified are not judges, but rather are scientists with special knowledge who are most familiar with the method or theory in question. Id. at 203. "Judicial acceptance of a scientific theory or instrument can occur only when it follows a general acceptance by the community of scientists involved." Id. at 196, quoting Commonwealth v. Fatalo, supra at 269. In determining whether general acceptance by the appropriate scientific community has occurred, "we may properly consider not only the testimony of experts in the record before us but also articles written by experts and the conclusions of other courts." Commonwealth v. Kater, 388 Mass. 519, 527 (1983). See Commonwealth v. Whynaught, 377 Mass. 14, 17-18 (1979).

We have applied the Frye rule not only in the polygraphy context, see Fatalo, supra, but in numerous other contexts as well. The rule is imbedded in our law. See, e.g., Commonwealth v. Gomes, 403 Mass. 258, 265-266, 270 (1988) (electrophoresis in blood-grouping analysis); Commonwealth v. Beausoleil, 397 Mass. 206, 215 (1986) (human leukocyte antigen [HLA] testing in paternity cases); Commonwealth v. Kater, supra at 531-534 (hypnotically aided testimony); Commonwealth v. Whynaught, supra at 18 (radar speedmeter); Commonwealth v. Lykus, supra at 205 (spectrographic voice analysis). Apparently, the Chief Justice would have us abandon that long-standing rule. We are not persuaded to do so.

More than a decade after our decision in Fatalo, which was based on our acceptance of the Frye rule, we deviated from the Frye rule. In Commonwealth v. A Juvenile, 365 Mass. 421, 429 (1974), we stated that, "despite very significant progress in recent years," polygraphy was still insufficiently reliable to satisfy the Fatalo requirements for general admissibility. Nevertheless, the court, with three Justices dissenting, concluded "that polygraph testing has advanced to the point where it could prove to be of significant value to the criminal trial process if its admissibility initially is limited to carefully defined circumstances designed to protect the proper and effective administration of criminal justice." Id. at 425. We subsequently observed in Commonwealth v. Vitello, 376 Mass. 426, 442 (1978), that "we recognized in A Juvenile that failure to achieve the standard of general acceptance need not freeze the evidentiary development of the polygraph in view of its unique potential as a tool of justice." So, in A Juvenile, supra at 432, instead of relying on the appropriate community of scientists to validate polygraph testing, we made our own evaluation of the polygraph's potential and announced a new rule as "a cautious first step toward the acceptance of polygraph testing." The new rule, to be applied only where the defendant moves the court for permission to submit to a polygraph examination, was stated as follows: "[I]f a defendant agrees in advance to the admission of the results of a polygraph test regardless of their outcome, the trial judge, after a close and searching inquiry into the qualifications of the examiner, the fitness of the defendant for such examination, and the methods utilized in conducting the tests, may, in the proper exercise of his discretion, admit the results, not as binding or conclusive evidence, but to be considered with all other evidence as to innocence or guilt. As a prerequisite the judge would first make sure that the defendant's constitutional rights are fully protected." A Juvenile, supra at 426. See id. at 430-431.

We have never determined that the appropriate scientific community, which includes physiologists and psychologists, has generally accepted the validity of polygraphy as a scientific means of detecting deception. In Commonwealth v. Vitello, supra at 431, we simply "accept[ed] as current and valid the finding of the court in A Juvenile that the `general acceptance' standard of Fatalo ha[d] not yet been achieved," and announced, id. at 453, that polygraph evidence "cannot be admitted as independent evidence of guilt or innocence," but that the polygraph examiner may "testify on the limited issue of the defendant's credibility as a witness," id. at 455. Thus, the evidence was stated to be admissible only if the defendant testifies and then only for corroboration or impeachment.

The extensive record in this case indicates that the theory and practice of polygraphy have not changed appreciably since our descriptions in A Juvenile, supra at 426-427, and Vitello, supra at 431-439. In brief, the polygrapher investigating a criminal incident usually employs the "control question technique." Under this technique, the examiner asks the subject "relevant" questions pertaining directly to the incident being investigated, and also "control" questions which deal with acts similar to the incident in question but are more general in nature. The control questions are designed to produce a strong physical reaction on the polygraph instrument, which records respiratory activity, sweat gland activity, and changes in blood pressure. The examiner asks several sequences of control and relevant questions, and then scores the result using one of several scoring methods. As a general matter, if control questions elicit stronger reactions than the relevant questions, the subject is considered to have answered the relevant question truthfully. If the relevant questions produce stronger responses, then the subject is considered to have been deceptive. See Raskin, The Polygraph in 1986: Scientific, Professional and Legal Issues Surrounding Application and Acceptance of Polygraph Evidence, 1986 Utah L. Rev. 29.

Now, fifteen years after our decision in A Juvenile, there remains no consensus among experts as to the accuracy of polygraph testing to detect deceit. One recent article cites figures of 97 per cent accuracy for guilty subjects and 92 per cent accuracy for innocent subjects, for an over-all rate of 95 per cent based on several laboratory studies "performed under carefully controlled conditions by highly skilled examiners with extensive psychological training and expertise." Raskin, supra at 42-43. This would give the polygraph test a probative value comparable to that which we found sufficient in admitting evidence from the HLA test on the issue of paternity in Commonwealth v. Beausoleil, 397 Mass. 206, 219 (1986). Another article concludes that scientifically credible studies show an average 84 per cent accuracy rate for guilty subjects and 53 per cent on innocent subjects. See Lykken, The Validity of Tests: Caveat Emptor, 27 Jurimetrics J. 263, 264 (Spring, 1987). Such a record gives the polygraph examiner only a slightly better chance of correctly identifying an innocent subject by using his machine than he would have by flipping a coin. See Lykken, The Lie Detector and the Law, 8 Crim. Def. 19, 26 (1981). A third recent study shows an accuracy rate of 74 per cent for guilty subjects and 72 per cent for innocent subjects. Kleinmuntz Szucko, A Field Study of the Fallibility of Polygraphic Lie Detection, 308 Nature 449 (1984). As this diversity of opinion shows, polygraphy has not met the "general acceptance" standard of Commonwealth v. Fatalo, 346 Mass. 266, 269 (1963).

Although experts disagree over the magnitude of the overall error rate to be expected from a polygraph test, they generally agree that the error rate for innocent subjects who take the test is higher than that for guilty subjects. The experts in this case agreed that the rate of innocents misidentified as guilty is roughly twice the rate of guilty subjects who pass the test. Even the most favorable of the above estimates of the test's accuracy suggests that eight per cent, or approximately one of every thirteen innocent defendants, will wrongly be evaluated as deceptive by a polygraph examiner.

As we have said earlier in this opinion, in determining whether a scientific method or theory has gained general acceptance among the relevant experts, we may properly consider the conclusions of other courts. Commonwealth v. Kater, 388 Mass. 519, 527 (1983). Commonwealth v. Whynaught, 377 Mass. 14, 18 (1979). Commonwealth v. Lykus, 367 Mass. 191, 199 (1975). There is suggestion in United States v. Piccinonna, 885 F.2d 166 (11th Cir. 1989), that the Court of Appeals for the Eleventh Circuit, by a divided court, has concluded that polygraphy has gained such acceptance. However, we are aware of no other court that has so concluded. Furthermore, in considering whether, in the absence of consensus among the appropriate scientific community, we should nevertheless admit polygraphic evidence in criminal trials, we are again assisted by knowing the course taken by other State and Federal courts. We discuss below the law elsewhere.

Numerous courts in other jurisdictions have either held or announced in dicta that polygraphic evidence is inadmissible even when the parties have entered into a pretest stipulation that the results will be admissible. Those courts generally have reasoned, as we did in Commonwealth v. Vitello, 376 Mass. 426, 448 (1978), that a prior stipulation cannot "`imbue' the polygraph test with reliability and probative value where such qualities are otherwise lacking." The courts of North Carolina, Oklahoma, and Wisconsin had previously admitted such evidence and then, on reexamination, decided not to do so. See State v. Grier, 307 N.C. 628 (1983); Fulton v. State, 541 P.2d 871 (Okla. Crim. App. 1975); State v. Dean, 103 Wis.2d 228 (1981). Still other jurisdictions in large number have dealt with cases in which, like here, there had been no stipulation and have held that polygraphic test results are inadmissible. Lastly, courts in other jurisdictions, dealing with cases in which the parties have entered into a pretest stipulation as to the admissibility of results, have sanctioned the admission of the evidence but have made clear that, in the absence of stipulation, the evidence would not have been admissible.

The cases cited in notes 4, 5, and 6, supra, most of which have been decided subsequent to our 1974 decision in Commonwealth v. A Juvenile, supra, demonstrate a nearly unanimous judicial consensus not only that polygraphy has not gained general acceptance among the community of scientists best qualified to judge its validity, but also that polygraphic evidence, at least in the absence of a pretest stipulation, ought not be admitted as evidence in a criminal trial for any purpose.

In the absence of a stipulation as to admissibility, our research discloses that, with few exceptions, polygraphic evidence is inadmissible throughout the country, even in the discretion of the trial judge, either for substantive purposes or to corroborate or impeach a witness. United States v. Piccinonna, supra, which holds that such evidence is admissible in the discretion of the trial judge, for impeachment or corroborative purposes, is an exception to that general rule. Only in New Mexico is polygraphic evidence admissible as a matter of right. See Tafoya v. Baca, 103 N.M. 56 (1985); State v. Dorsey, 88 N.M. 184 (1975). We note that there is no Frye-type ( Frye v. United States, 298 F. 1013, 1014 [D.C. Cir. 1923]) analysis in either New Mexico opinion.

In developing our approach to the use of polygraphic evidence in A Juvenile and Vitello, we were keenly aware of the shortcomings of the polygraph method of ascertaining deception, and its use in a trial situation. We noted the subjective nature of the polygraph method, and the crucial role played by the "competence, experience, and education of the test examiner." A Juvenile, supra at 427. Vitello, supra at 438-439. We reiterated our concerns about the "uncertain reliability of polygraph evidence," Vitello, supra at 442, and "the danger that on the introduction of such evidence a trial could descend into a battle of experts on the probative value of the polygraph test rather than a determination of the guilt or innocence of a defendant." Id. at 442, quoting Fatalo, supra at 269. We noted the potential for confusing and prejudicing the jury and the related possibility that the use of polygraph evidence "may usurp the jury's historic role of determining the credibility of witnesses, and finding facts." Vitello, supra at 445. See A Juvenile, supra at 447 (Quirico, J., dissenting). We also addressed the burden placed on trial judges by the need to determine in each case whether the expert is fully qualified and the test was properly conducted. Vitello, supra at 446-447. Nonetheless, we took a "cautious first step" in the hope, if not expectation, that the development of polygraphic testing would soon reach the stage where it would be generally accepted by the appropriate scientific community. A Juvenile, supra at 432. We considered "the time [to be] ripe for cautious judicial examination and evaluation," but we recognized that "[t]he experience gained may well lead to a total rejection of the concept." Id. at 434.

Fifteen years has been more than enough time for examination and evaluation. As is apparent from the judge's memorandum, the evidentiary shortcomings of polygraphy have not been alleviated in the slightest way. In addition, it is clear from the extensive record in this case and the available literature that our hope that polygraphy would mature to the point of general scientific acceptance has not materialized. Further hope or expectation in that regard is no longer warranted. Thus, whatever justification there may have been for our single departure from the Frye rule in A Juvenile and Vitello, that justification no longer exists. Accordingly, supported by the overwhelming authority throughout the country, we announce that polygraphic evidence, with or without pretest stipulation, is inadmissible in criminal trials in this Commonwealth either for substantive purposes or for corroboration or impeachment of testimony. The rule we adopt today governs this case and all other cases in which the trial begins after this decision.

The order allowing the defendants' motions for court-ordered polygraph tests is vacated.

So ordered.