In this case we must decide the meaning of the word “likely” as the Legislature used it in the context of defining a sexually dangerous person as someone “likely to engage in sexual offenses if not confined to a secure facility” pursuant to G. L. c. 123A, § 1.
Background. In December, 1987, Steven Boucher pleaded *275guilty to three counts of rape of a child with force and three counts of indecent assault and battery on a child under the age of fourteen years. He was given three sentences of from twelve to fifteen years and three sentences of from eight to ten years, all to be served concurrently. Shortly before Boucher was to complete his sentences and be released from custody, the Commonwealth filed a petition for his commitment as a sexually dangerous person pursuant to G. L. c. 123A, §§ 1, 12, contending that he met the statutory requirements of a “person who has been . . . convicted ... of a sexual offense and who suffers from a mental abnormality or personality disorder which makes the person likely to engage in sexual offenses if not confined to a secure facility” (emphasis added). G. L. c. 123A, § 1.
Following a two-day jury-waived trial in the Superior Court, the trial judge found beyond a reasonable doubt that Boucher (1) had been convicted of a sexual offense; (2) suffered from a “mental abnormality” (pedophilia) as a result of which he had serious difficulty in controlling his behavior; and (3) was predisposed “to committing sexual offenses to such an extent that there is a significant risk of his committing another sexual offense.” The judge further found that the risk of Boucher’s committing additional sexual offenses was sufficient “to make [him] a menace to the safety of others.” The judge dismissed the petition, however, because he concluded that the Commonwealth had failed to prove beyond a reasonable doubt that Boucher was “likely” to commit new sexual offenses unless confined to a secure facility. He based this conclusion on his interpretation of the term “likely” in the statute to mean “more likely than not,” or, stated otherwise, that the Commonwealth had the burden of proving that the risk of Boucher committing an additional sexual offense was “at least fifty percent” (emphasis in original). The Commonwealth appealed, and the dismissal was stayed by a single justice of the Appeals Court pending the outcome of the appeal. We granted the defendant’s application for direct appellate review.1
Discussion. With respect to the weighing and crediting of *276testimony admitted at trial, we will not substitute our judgment for that of the trier of fact. Hill, petitioner, 422 Mass. 147, 156 (1996). We do, however, scrutinize without deference the propriety of the legal criteria employed by the trial judge and the manner in which those criteria were applied to the facts. Id. In this case, the judge required the Commonwealth to prove that Boucher was “more likely than not” to commit a sexual offense as the result of his mental abnormality when the plain language of the statute required only that Boucher be “likely” to commit such an offense. Because we conclude that these terms have different meanings, the judge erred by applying the wrong legal criteria.
“Likely” is not defined in G. L. c. 123A, nor is it understood to be a term of art requiring a specific and limited interpretation. Where a statutory term is undefined, it must be understood in accordance with its generally accepted plain meaning. Commonwealth v. Brown, 431 Mass. 772, 775 (2000). Commonwealth v. Campbell, 415 Mass. 697, 700 (1993). Dictionaries define “likely” as “seeming as if it would happen,” “reasonably to be expected,” “apparently destined,” and “of such a nature or so circumstanced as to make something probable.” Webster’s New World Dictionary 891 (2d ed. 1970). Webster’s Third New Int’l Dictionary 1310 (1993). As commonly used and understood, “likely” is a word that encompasses a range of probabilities depending on the specific context in which it is used. We conclude that something is “likely” if it is reasonably to be expected in the context of the particular facts and circumstances at hand.
In assessing the risk of reoffending, it is for the fact finder to determine what is “likely.” Such a determination must be made on a case-by-case basis, by analyzing a number of factors, including the seriousness of the threatened harm, the relative certainty of the anticipated harm, and the possibility of successful intervention to prevent that harm. See, e.g., Cross v. Harris, 418 F.2d 1095, 1100-1101 (D.C. Cir. 1969) (“It may well be impossible to provide a precise definition of ‘likely’ as the term is used in the [D.C. sexual psychopath] statute. The degree of *277likelihood necessary to support commitment may depend on many factors”). While “likely” indicates more than a mere propensity or possibility, it is not bound to the statistical probability inherent in a definition such as “more likely than not,” and the terms are not interchangeable. To conclude that “likely” amounts to a quantifiable probability, absent a more specific statutory expression of such a quantity, is to require mathematical precision from a term that, by its plain meaning, demands contextual, not statistical, analysis.2
“More likely than not” is most familiar as a standard of proof equivalent to fifty per cent plus one. As used in the statute, however, the term “likely” is not intended as a standard or burden of proof. Rather, it is descriptive of one characteristic (“likely to engage in sexual offenses”) of a sexually dangerous person. While the Commonwealth is required to prove beyond a reasonable doubt that a person is sexually dangerous, that is, has all the characteristics of such a person as defined in G. L. c. 123A, § 1, it is not required to prove to any particular mathematical quantum the likelihood of his committing another sexual offense.
In analogous circumstances, the United States Supreme Court reached a similar result regarding the constitutional requirements of the Kansas Sexually Violent Predator Act. In Kansas v. Hendricks, 521 U.S. 346 (1997), the Court held that commitment as a sexually dangerous person is only appropriate where, among other prerequisites, a person is shown to “suffer from a volitional impairment rendering [him] dangerous beyond [his] control.” Id. at 358. In Kansas v. Crane, 534 U.S. 407 (2002), the Court addressed the issue of what degree of “inability to control” one’s behavior the government was required to establish in order to prove this necessary characteristic. Id. at 413. It held that the inability to control behavior need not be demonstrated with mathematical precision, and that “[i]t is *278enough to say that there must be proof of serious difficulty in controlling behavior.” Id. The Court further concluded that the adequacy of this proof must be viewed in the context of the features of the case (e.g., the nature and seriousness of the abnormality) and, most importantly, when so viewed it “must be sufficient to distinguish the dangerous sexual offender whose . . . mental illness [or] abnormality . . . subjects him to civil commitment from the dangerous but typical recidivist convicted in an ordinary criminal case.” Id. There is little question that, in the circumstances of our statute, proof that a person is likely to commit another sexual offense need not be established to a mathematical certainty above fifty per cent in order to distinguish the dangerous sexual offender from “the dangerous but typical recidivist convicted in an ordinary criminal case.”3
In addition, there is nothing in the statute or its history to suggest that the Legislature intended that “likely” be interpreted to mean “more likely than not.” As a matter of legislative intent, such an interpretation would be incongruous given the unmitigated use of the word “likely” in the statute. If the Legislature had intended to restrict “likely” to a usage less broad than its commonly understood meaning, the Legislature would have done so. For example, G. L. c. 123, § 7, provides that “the superintendent of a [mental health] facility may petition the district court ... for the commitment . . . and retention of any patient . . . whom said superintendent determines that the failure to hospitalize would create a likelihood of seri*279ous harm by reason of mental illness” (emphasis added). Despite the similarity in tenor and effect between G. L. c. 123 and G. L. c. 123A, there is a stark contrast — G. L. c. 123 provides a specific statutory definition of “[l]ikelihood of serious harm” that varies from a “substantial risk” to a “very substantial risk” depending on the nature of the perceived harm.4 In other statutes, as well, the Legislature has qualified the generally understood broad scope of “likely” by including limiting language. See, e.g., G. L. c. 21E, § 2 (requiring “substantial likelihood” of release of oil or hazardous materials to establish a “[tjhreat of release”); G. L. c. 71B, § 3 (allowing student to be removed from school if school committee shows that student’s behavior poses “substantial likelihood of injury to himself or others”); G. L. c. 209A, § 4 (allowing courts to issue temporary relief orders in domestic abuse cases where “plaintiff demonstrates a substantial likelihood of immediate danger of abuse”).
By contrast, G. L. c. 123A, § 1, only requires a determination that a person is “likely to engage in sexual offenses.” The lack of restrictive language around “likely” in c. 123A, § 1, indicates the legislative intent to employ the term fully in accordance with its commonly understood meaning.5
Further evidence of legislative intent can be gleaned from *280looking at how the term “likely” was interpreted in G. L. c. 123A before that statute was amended in 1999. St. 1999, c. 74, § 6. Prior to its amendment, G. L. c. 123A, § 1, required a finding that the person was “likely to attack or otherwise inflict injury on the objects of his uncontrolled or uncontrollable desires.” G. L. c. 123A, § 1, as appearing in St. 1993, c. 489, § 1. As in the current version of G. L. c. 123A, the previous version contained neither a statutory definition of “likely” nor restrictive language limiting the term to a use other than its common understanding. Because the use of “likely” in the current and prior versions of G. L. c. 123A, § 1, is so similar, and given the well-settled and logical mle of statutory construction that, when the Legislature reenacts a statute using the same language, the Legislature must be presumed to have adopted the prior judicial' construction of that language, we can look for guidance to judicial interpretations of the term in its prior incarnation. Boston Hous. Auth. v. Bell, 428 Mass. 108, 110 (1998); Commonwealth v. Miller, 385 Mass. 521, 524 (1982).
In several cases decided under the prior version of G. L. c. 123A, § 1, this court reviewed whether there was sufficient evidence adduced at trial to support a finding that the defendant was “likely” to commit an attack or inflict injury. The holdings *281in those cases are consistent with the broad common understanding of the term “likely” as compared to the more limited understanding pressed by the defendant. See, e.g., Commonwealth v. Rodriguez, 376 Mass. 632, 643-645 (1978) (upholding lower court finding that defendant was sexually dangerous person where witnesses agreed that there was “significant probability, hard to reduce to a percentage figure, that he would commit further violent sexual crimes”); Commonwealth v. Walsh, 376 Mass. 53, 59 & n.3 (1978) (finding “sufficient evidence on the issue whether the defendant is likely to commit future violent [sexual] acts” where there was testimony that “there is a good possibility” that defendant will recidivate); Commonwealth v. McHoul, 372 Mass. 11, 14 (1977) (upholding lower court finding that defendant was sexually dangerous person where there was testimony that “there was a strong propensity for repetition of the defendant’s past acts . . . that ‘there is a likelihood that there will be future violent behavior’ and that there was ‘reason to expect that [the defendant] would repeat some sort of violent sexual crime’ ”); Commonwealth v. Dagle, 345 Mass. 539, 543 (1963) (upholding lower court finding that defendant was sexually dangerous person where there was testimony that there was “distinct possibility” that defendant would commit other sexual offenses if released). These decisions do not suggest that “likely” meant “more likely than not” in the prior version of G. L. c. 123A, § 1, and so we presume that the Legislature did not intend “likely” to be so construed when it amended the statute but retained the term.
Conclusion. For all the reasons stated above, we find that the judge erred by interpreting the term “likely” to mean “more likely than not,” and applying that erroneous interpretation to the evidence at trial. Therefore, the order of the Superior Court dismissing the petition is vacated and the case is remanded to the trial judge to determine whether the evidence, in the context of the particular facts and circumstances of the case, supports a finding that Boucher would reasonably be expected to engage in sexual offenses if not confined to a secure facility.
So ordered.
In a footnote to his findings of fact, rulings of law, and order, the trial judge noted that, for the purposes of this case, the parties had agreed that the term “likely” meant “more likely than not.” This does not appear to be fully *276supported in the record, and at oral argument the parties agreed that the Commonwealth had not conceded the issue at trial.
Such a contextual construction of “likely” is not so vague as to be unfair. A law is not unduly vague just because it calls for the application of “an imprecise but comprehensible normative standard.” Commonwealth v. Gallant, 373 Mass. 577, 580 (1977), quoting Commonwealth v. Orlando, 371 Mass. 732, 734 (1977). It is enough if the term in question has a commonly understood meaning. Commonwealth v. Arthur, 420 Mass. 535, 539 (1995), and cases cited.
Our conclusion that the term “likely” does not mean “more likely than not” is consistent with a recent decision of the Supreme Court of California, interpreting its own Sexually Violent Predators Act. People v. Superior Court, 27 Cal. 4th 888, 895 (2002). Similar to our statute, the California act requires the government to prove that an individual “has a diagnosed mental disorder so that he or she is likely to engage in acts of sexual violence without appropriate treatment and custody” (emphasis added). Cal. Welf. & Inst. Code § 6601(d) (West 1998 & Supp. 2002). The court, when asked to determine whether the use of the word “likely” required the fact finder to determine that there was a “better than even chance of new criminal sexual violence,” concluded that it did not. People v. Superior Court, supra. Rather, the court held that this characteristic is proved on a showing that the person presents a “substantial danger — that is, a serious and well-founded risk — of criminal sexual violence.” Id. If such a risk is present, it is legally erroneous to hold that the requirement is not met just because the evidence does not establish that the risk exceeds fifty per cent. Id.
“Likelihood of serious harm” is defined in G. L. c. 123, § 1, to mean: “(1) a substantial risk of physical harm to the person himself as manifested by evidence of, threats of, or attempts at, suicide or serious bodily harm; (2) a substantial risk of physical harm to other persons as manifested by evidence of homicidal or other violent behavior or evidence that others are placed in reasonable fear of violent behavior and serious physical harm to them; or (3) a very substantial risk of physical impairment or injury to the person himself as manifested by evidence that such person’s judgment is so affected that he is unable to protect himself in the community and that reasonable provision for his protection is not available in the community” (emphases added).
Many other Massachusetts statutes use the term “likely” in the broad and unfettered manner employed by G. L. c. 123A. E.g., G. L. c. 21E, § 2 (in context of Oil and Hazardous Material Release Prevention and Response Act, “[condition of substantial release migration” is defined as “release of oil or hazardous material that is likely to be transported through environmental media where the mechanism, rate or extent of transport has resulted in or, if not promptly addressed, has the potential to result in: [a] health damage, safety hazards or environmental harm; or [6] a substantial increase in the extent or magnitude of the release, the degree or complexity of future response actions, or the amount of response costs” [emphasis added]); G. L. c. 23D, *280§ 13 (any business may seek financing from economic stabilization trust if, among other requirements, “the business is likely to experience a large loss of employment, plant closure, or failure without the loan financing or investment by the trust” [emphasis added]); G. L. c. 38, § 6 (requiring notification of State archeologist if “skeletal remains are deemed likely to be Native American” [emphasis added]); G. L. c. 132, § 13 (allowing destruction of trees infested with beetles that spread Dutch elm disease, or are “likely to become so infested" [emphasis added]); G. L. c. 268A, § 23 (b) (3) (precluding public employees from acting “in a manner which would cause a reasonable person ... to conclude . . . that he is likely to act or fail to act as a result of kinship, rank, position or undue influence of any party or person” [emphasis added]); G. L. c. 269, § 17 (defining criminal hazing to include “any other brutal treatment or forced physical activity which is likely to adversely affect the physical health or safety of any such student or other person, or which subjects such student or other person to extreme mental stress” [emphasis added]). We do not see any legislative intent to adopt a one-dimensional definition of a commonly understood term that is used broadly in so many different statutes. What is “likely” pursuant to any particular statute requires some understanding of the purpose of the statute, the nature of the perceived harm, and the context in which the term is used. The Legislature intended a contextual analysis, not a strict mathematical definition.