dissenting. Today, the Court holds that, in cases in which a sentence of life imprisonment may be imposed, the presumption that an accused should remain at liberty pending trial is “reversed” so that the presumption becomes that the accused should be incarcerated. This extraordinary proposition finds no support in our precedents or in those of the United States Supreme Court. Nor is it compelled by, or even a logical conclusion from, Vermont’s constitutional and statutory provisions that, in capital or life imprisonment cases, where evidence of guilt is great, the defendant shall not be bailable as a matter of right. On this ground alone, I am compelled to dissent.
Regrettably, the Court compounds its error by sanctioning a bail decision-making process that is both entirely without standards and completely unreviewable by this Court. I do not think that such a holding comports with due process or is consistent with this Court’s responsibility to provide appellate review of trial court decisions.
I.
Under United States v. Salerno, 481 U.S. 739 (1987), both substantive and procedural due process safeguards protect bail decision making. Substantive due process protects an individual from government interference with “rights ‘implicit in the concept of ordered liberty.’” Id. at 746 (quoting Palko v. Connecticut, 302 U.S. 319, 325-26 (1937)). Procedural due process guarantees that any permissible governmental interference ■with a right is done in a fair manner. Id.
Both due process analyses begin with an examination of the individual interest threatened by governmental action. Under substantive due process, if governmental action threatens a right deemed “fundamental,” then courts “must examine carefully the importance of the governmental interests advanced and the extent to which they are served by the challenged regu*465lation,” Moore v. East Cleveland, 431 U.S. 494, 499 (1977), and uphold the government action only if it is narrowly tailored to further a “legitimate and compelling” governmental interest. Salerno, 481 U.S. at 749-50. Procedural due process imposes constraints on any governmental decision that would deprive an individual of liberty or property, Mathews v. Eldridge, 424 U.S. 319, 332 (1976)), requiring that any procedure used to effectuate such a deprivation be analyzed by balancing the protected interest against any governmental interests, taking into account the risk of erroneous deprivation. Id. at 335.
Thus, both substantive and procedural due process apply to bail decisions, not because of any constitutional right to bail, but because liberty is a fundamental right independently guaranteed. “Freedom from bodily restraint has always been at the core of the liberty protected by the Due Process Clause from arbitrary governmental action.” Foucha v. Louisiana, — U.S. —, —, 112 S. Ct. 1780, 1785 (1992). See also Aime v. Commonwealth, 611 N.E.2d 204, 210-11 (Mass. 1993) (tracing central importance of liberty in Anglo-American common law and constitutional law and concluding that freedom from governmental restraint “lies at the heart of our system of government and is undoubtedly a fundamental right”).
Those charged with crimes who come before the court for bail decisions are not criminals. They are merely accused, not convicted; they enjoy a right to liberty by virtue of the presumption of innocence. “The principle that there is a presumption of innocence in favor of the accused is the undoubted law, axiomatic and elementary, and its enforcement lies at the foundation of the administration of our criminal law.” Coffin v. United States, 156 U.S. 432,453 (1895). The presumption cannot be legislated away; it is part of the right to fair trial protected by the Fourteenth Amendment. Estelle v. Williams, 425 U.S. 501, 503 (1976). Consequently, those accused of crimes are on the same legal footing as other noncriminal persons facing possible deprivations of liberty. Although governmental needs may outweigh an individual’s liberty interest, Salerno, 481 U.S. at 748, liberty, even for the accused, “is the norm, and detention prior to trial or without trial is the carefully limited exception.” Id. at 755. See also State v. Duff, 151 Vt. 433, 440, 563 A.2d 258, 263 (1989) (pretrial detention “necessarily cuts against the pre*466sumption of innocence inherent in our criminal jurisprudence” and is therefore limited).
Instead of guaranteeing that detention will be a “carefully limited exception,” the majority begins with the dubious proposition that if defendant has no constitutional right to bail, the presumption in favor of liberty is somehow “switched so that the norm is incarceration and not release.” Neither Salerno nor any other United States Supreme Court case supports this presumption. Nor do I find support for it in Duff or In re Dexter, 93 Vt. 304, 107 A. 134 (1919), the two Vermont cases cited by the majority. Even if there were support in the later two cases, I would discount it. They are not due process cases, and whether the court abused its discretion in denying bail can only be meaningfully examined in a due process context. Salerno provides that context.
Under Salerno, the initial due process burden is on the State. To overcome the accused’s right to liberty, it must show a “legitimate and compelling” governmental need. Preventive detention for the safety of the community is the most frequently recognized need. See Ludecke v. Watkins, 335 U.S. 160, 171-72 (1948) (approving unreviewable executive power to detain enemy aliens during war); Carlson v. Landon, 342 U.S. 524, 537-42 (1952) (detention of Communist resident aliens pending deportation proceedings); Addington v. Texas, 441 U.S. 418, 426 (1979) (detention of dangerous mentally unstable individuals); Salerno, 481 U.S. at 750 (detention of dangerous criminal defendants). Other compelling interests recognized in the criminal context are risk of flight and danger to witnesses. Id. at 749.
Nothing in the Vermont Constitution or bail statutes explicitly states a compelling governmental interest to be served by denying bail. Chapter II, § 40 of the Vermont Constitution provides:
All persons, unless sentenced, or unless committed for offenses punishable by death or life imprisonment when the evidence of guilt is great, shall be bailable by sufficient' sureties. Persons committed for offenses punishable by death or life imprisonment, when the evidence of guilt is great, shall not be bailable as a matter of right.
This provision only identifies a class of people who are not bailable as a matter of right — those subject to penalties of death or *467life imprisonment where evidence of guilt is great. The relevant statute, 13 V.S.A. § 7553, tracks the constitution:
A person charged with an offense punishable by life imprisonment when the evidence of guilt is great shall not be bailable as a matter of right. If the evidence of guilt is not great, the person shall be bailable in accordance with section 7554 of this title.
Thus, § 7554 guides the trial court’s decision-making process in all bail cases except when the accused has no right to bail and evidence of guilt is great. In the latter cases, bail is left to the trial court’s discretion. Duff, 151 Vt. at 441, 563 A.2d at 263.
The bail amendment and corresponding statute manifest two governmental interests. Seriousness of potential punishment is a surrogate for seriousness of the crime; great evidence of guilt of the charged crime is a surrogate for likelihood of conviction. The only persons not bailable as a matter of right are those who meet both criteria. This combination suggests that the strongest governmental interest manifest in our bail scheme is potential for flight: those most likely to be convicted of crimes carrying the most severe punishments are the most likely to flee. Preventing danger to the public is a secondary — and, for purposes of due process balancing, a weaker — interest. If potential dangerousness were the primary interest, then no one accused of a crime, who demonstrated a propensity toward violence, would be bailable as of right. Legislative history of the 1982 bail amendment, the current version of § 40, shows that such a broad approach to dangerousness was specifically rejected. State v. Sauve, 159 Vt. 566, 572, 621 A.2d 1296, 1300 (1993).
Detention for violations of prior conditions — the reason relied on by the trial court and affirmed by the majority — is neither explicitly nor implicitly a governmental interest to be vindicated under our bail scheme. See id. at 575, 621 A.2d at 1301 (bail revocation for repeated violations of conditions does not rise to the level of a compelling governmental interest). Such violations should be relevant only to the extent that they bear on governmental interests that are part of the bail scheme — preventing flight, and to a lesser extent, minimizing danger to the public. The trial court did not find defendant was at risk to flee or endanger the public. Likewise, it did not con-*468elude that his prior violations rendered him a greater risk to flee or endanger the public. Its only conclusion was that “defendant is simply not motivated to abide by orders of the court.” This reasoning is far too broad to satisfy due process.
Moreover, even if we accept that our current bail scheme has, as a governmental interest, preventing danger to the public, I do not understand how the majority moves from recognizing that interest to creating a presumption of incarceration for all those without a right to bail. We are not free to interpret our bail provisions — statutory or constitutional — in a way that violates the fundamental right to liberty guaranteed by the Fourteenth Amendment to the United States Constitution. Hence, we cannot meaningfully analyze § 40 in isolation from the requirements of the due process clause as discussed in Salerno. Salerno allows for a presumption of dangerousness, if appropriately limited, but the legislature has simply not created one here. Rather, it merely gave courts the discretion to deny bail to some defendants, and Salerno limits how courts can exercise that discretion. Under Salerno, even a presumption of dangerousness is only the beginning of a carefully limited inquiry into whether the accused is in fact dangerous. And even if the accused is found — not presumed — dangerous, the court must still inquire whether incarceration is necessary to prevent the danger. The only presumption that I find any support for is the presumption of innocence.
II.
Assuming that our bail scheme manifests a legitimate governmental interest, it is still inadequate under Salerno. Due process requires that a bail decision be narrowly tailored to effectuate a compelling governmental interest. Here, no statutory factors guide the trial court’s discretion. Neither today’s opinion nor prior Vermont case law provides any such guidelines. Instead, the majority relies on Dexter, 93 Vt. at 315, 107 A. at 138, for the proposition that if the exercise of discretion is “controlled by certain and well defined and established rules,” then “we would leave the matter as there determined, let it be either way, without further consideration.” This would be an unremarkable proposition except the “well defined and established rules” are not articulated in Dexter or any other case. *469Apparently the majority believes that as long as the trial court follows unstated rules, this Court must rubber stamp the result. This is not the strict scrutiny required by due process; it is no scrutiny at all.
The closest this Court has come to making “well defined and established rules” is Duff, which suggests that the trial court look at considerations similar to those used in determining post-conviction bail. 151 Vt. at 441, 563 A.2d at 263-64 (citing with apparent approval Fountaine v. Mullen, 336 A.2d 1138, 1144 (R.1.1976)). Today, the Court, while apparently citing Duff with approval, takes a step back from that decision by failing to apply any such standards. Post-conviction bail decision making is controlled by statute. See 13 V.S.A. § 7574 (judge shall consider factors set forth in 13 Y.S.A. § 7554(b) and defendant’s conduct during trial and the fact of conviction) and 13 V.S.A. § 7554(b) (judge shall look at “nature and circumstances of the offense charged, the weight of the evidence against the accused, the accused’s family ties, employment, financial resources, character and mental condition, the length of residence in the community, record of convictions, and record of appearance at court proceedings or of flight to avoid prosecution or failure to appear at court proceedings”; judge may also consider any “[rjecent history of actual violence or threats of violence” as bearing on the “character and mental condition of the accused”). The majority does not adopt these or any such guiding considerations for pretrial bail decisions. We are left with an irrational result: persons merely accused of crimes, who still enjoy a presumption of innocence, are given less due process protection than those convicted of crimes, who are no longer entitled to that presumption.
The legislature, of course, could pass a statute setting standards for release of those not entitled to bail. It has not done so and need not do so, but, in the absence of such a statute, due process requirements apply no less rigorously. This Court must either set standards for principled bail decision making or require that the trial court set and apply such standards. Neither was done here.
In Salerno, the Court upheld the federal bail act because it found the act appropriately balanced the government’s need to protect the community from dangerous defendants with nu*470merous provisions protecting defendants. See 481 U.S. at 751. Clarification of what due process requires in denial-of-bail decisions is provided in Aime, 611 N.E.2d 204, and Witt v. Moran, 572 A.2d 261 (R.I. 1990).
In Aime, the Massachusetts Supreme Judicial Court struck down, on due process grounds, recently enacted amendments to its state’s bail law, which provided that a judge may exercise discretion to refuse bail to any person whose release will endanger the safety of any person or the community. 611 N.E.2d at 206. Analyzing the Massachusetts statute under Salerno, the court held that the state could not enact a preventive detention scheme “without providing safeguards similar to those which Congress incorporated into the Bail Reform Act.” Id. at 212. The court found the bail amendments fatally defective, in many ways not applicable here, but was especially critical of the “unbridled discretion” given to judges “to determine whether an arrested individual is dangerous.” Id. at 214. In contrast, the federal bail statute contains provisions that, consistent with substantive due process, narrow the judge’s inquiry (for example, the court must find that no release conditions “will reasonably assure the . . . safety of any other person and the community,” 18 U.S.C. § 3142(f)), and the statute enumerates factors to be considered in the bail decision.
Similarly, the Rhode Island Supreme Court applied Salerno to its own state’s bail statute, which allows pretrial detention of certain classes of defendants defined by statute as dangerous. The court upheld the statute but also formulated procedures necessary to make it conform to federal due process, including a requirement that, in deciding whether to deny bail, the trial court make findings of fact on the record regarding the individual defendant’s dangerousness. Witt, 572 A.2d at 266. Next, the court held that any restrictions on defendant’s liberty must be consistent with the particular circumstances of the case. Id. at 267. To guarantee this, the trial court must consider any alternatives to denying bail that would ensure that the defendant would not be a danger to the community. Id. Finally, the court endorsed Salerno’s procedural due process protections, including the clear-and-convincing standard for a finding of dangerousness, reasoning that a heightened standard promotes a more accurate judicial determination, which is necessary to justify restrictions on defendant’s liberty prior to trial. Id.
*471I believe Aime and Witt properly apply due process standards mandated by Salerno to bail review determinations for defendants who are not guaranteed bail as of right. Like the statute analyzed in Witt, our bail statute, 13 V.S.A. § 7553, singles out a class of offenders who may be denied bail. But merely being a member of the class is not enough to support preventive detention. Rather, the court must exercise its discretion within the context of an individualized inquiry outlined in Witt and find by clear and convincing evidence that a particular defendant would be a danger to an individual or the community and that no conditions of release would be sufficient to prevent that danger. I would remand for proceedings consistent with that purpose. See Duff, 151 Vt. at 442, 563 A.2d at 264 (where trial court made inadequate findings on bail decision, case was remanded so it could show on the record how it had exercised its discretion).
On remand, I would also require that the trial court appropriately apply the clear and convincing standard. Having improperly invented a presumption of incarceration, the Court puts the burden of proof on defendant to show that he deserves liberty. For example, the Court states that “it is entirely appropriate for the court to deny bail unless it is fully convinced that the defendant will abide by the conditions that would be imposed if defendant were released,” even though it apparently accepts the Salerno requirement that there must be clear and convincing evidence that defendant is a risk to flee or endanger the public in order to deny bail. Cf. Foucha, — U.S. at —, 112 S. Ct. at 1787 (striking, on due process grounds, a Louisiana statute requiring persons to prove that they are not dangerous in order to be freed from confinement in a mental institution).
The Court states that Salerno has “only limited applicability” because it upholds the federal bail statute without making a definitive ruling on what parts of it are constitutionally required. Witt and Aime, however, treat all of the Salerno protections as due process requirements. Also, the Court seeks to distinguish all three cases because they deal with bail statutes that are explicitly preventive detention measures. I find this a distinction without a difference. Surely 13 V.S.A. § 7553, at least as the Court today interprets the bail scheme, provides for preventive detention. Surely that was *472what the trial court sought to achieve in this case. What is lacking in the Court’s interpretation of our bail scheme is not the goal of preventive detention, but any reasoned limitations on the court’s discretion to detain an accused who lacks a constitutional or statutory right to bail.
III.
Finally, and most disturbing from a due process perspective, is the Court’s approach to reviewing the trial court’s bail decision. Review in this Court should, as the Court today asserts, “provideQ the necessary check against unbridled discretion.” Because defendant’s fundamental right to liberty is at stake and because any attempt to deprive him of liberty should be strictly scrutinized, the Court’s review of the trial court’s discretion falls woefully short. Instead of requiring the trial court to make a record that conforms to due process, the Court simply hypothesizes the trial court’s reasoning and allows it the benefit of various presumptions — for example, a presumption that defendant’s “‘history of pretrial criminality’” leads to the conclusion that “‘he cannot be trusted to conform to the requirements of the law while on release,”’ 160 Vt. at 462, 631 A.2d at 1141 (quoting United States v. Coyne, 800 F. Supp. 1018, 1020 (D. Mass. 1992)), which the Court then enlarges to a presumption that multiple violations of prior conditions lead to a conclusion that “no combination of conditions of release would ensure defendant’s presence at trial or the safety of the alleged victim and witnesses and the public.” Id. This use of unsubstantiated presumptions is inconsistent with the due process requirement that decisions involving fundamental rights must be narrowly and individually tailored. Bail inquiries should be based on facts, not presumptions or speculations: Was this defendant at risk to flee or endanger others? Who was endangered — the victim, witnesses, the public? Was there any condition that would effectively protect the endangered person^)? If not, why not?
This case illustrates the danger of allowing a looser inquiry. Defendant came before the court for a review of a no-bail decision on a charge of aggravated sexual assault. Reading the transcript and the court’s opinion, the court apparently looked at defendant’s prior history in the court file, found violations of *473conditions and charges of violations of conditions, and decided that defendant could not comply with conditions and should be incarcerated. The issue is not, as the Court states, whether the trial court can in its discretion deny bail for “substantial” violations of “important” conditions of release. These characterizations are the Court’s. The trial court never did any such analysis.
The trial court did not substantively evaluate defendant’s violations of conditions. If the court found the violations “substantial” or “serious,” it did not say so. The prior violations, all apparently involving defendant illegally approaching a former girlfriend, were not related to the current charge. Under Sauve, defendant could not have his bail revoked for any of these violations or for violating the no-alcohol condition. 159 Vt. at 575-76, 621 A.2d at 1302. The court provides no explanation about how the former violations relate to the new charge. Finding prior violations in the record should be the beginning not the end of the inquiry.
Salerno requires that pretrial detention be a “carefully limited exception.” 481 U.S. at 755. The Court today does not require the trial court to be either careful or limited in its decision making; it has reduced a fundamental concept to an empty formula. Under the Court’s decision, an accused could be denied liberty merely because it is inconvenient for the court system to deal with him, as long as some hypothetical governmental interest can be produced as a cover.
Moreover, the problem with the trial court decision here is not that it is wrong, but that we cannot tell if it is wrong. The trial court’s rationale is accepted because it “is not arbitrary.” Both the trial court’s decision making and this Court’s review of it fall far short of the strict scrutiny required under substantive due process analysis.
IV.
The biggest loser in today’s decision is not the bail decision-making process. It is the presumption of innocence. Based only on an accusation — not a conviction — of aggravated sexual assault, defendant has lost his presumption of innocence and faces indefinite incarceration, without a trial, without even a finding that he is at risk to flee or endanger others.
*474Although the Court stresses that bail as of right is denied only to those committing the “most serious of crimes,” the legislature can easily enlarge that class by attaching the penalty of possible life imprisonment to any crime. The presumption of innocence is too easily eroded by fear of those accused of criminal behavior. We should not allow that erosion to be compounded by allowing the process of making bail decisions to be a casual one.
Honoring the presumption of innocence is often difficult; sometimes we must pay substantial social costs as a result of our commitment to the values we espouse. But at the end of the day the presumption of innocence protects the innocent; the shortcuts we take with those whom we believe to be guilty injure only those wrongfully accused and, ultimately, ourselves.
Salerno, 481 U.S. at 767 (Marshall, J., dissenting).
I would reverse and remand for a bail review that comports with the requirements of due process.
I am authorized to state that Justice Morse joins in this dissent.