OPINION ON STATE’S PETITION FOR DISCRETIONARY REVIEW
MALONEY, Judge.Appellant was convicted by a jury on an indictment charging the offense of Unauthorized Use of a Motor Vehicle, V.T.C.A. Penal Code, § 31.07. The indictment contained enhancement paragraphs charging two prior felony convictions; punishment was assessed by the jury at 40 years’ confinement in the Texas Department of Criminal Justice, Institutional Division. On appeal, in an unpublished opinion, the Court of Appeals reversed the conviction on the basis of insufficient evidence and remanded the case to the trial court to enter a judgment of acquittal. Geesa v. State, No. 2-88-140-CR (delivered February 21, 1990).
This Court granted the State’s petition for discretionary review to determine (1) whether this Court should reject the “reasonable-hypothesis-of-innocence analytical construct” utilized in review of circumstantial evidence cases; and (2) whether the Court of Appeals erred in finding the evidence insufficient in this case.
We find (1) that the analytical construct is not valid in light of this Court’s earlier decision to abrogate the circumstantial evidence charge, see Hankins v. State, 646 S.W.2d 191 (Tex.Cr.App.1983) (opinion on rehearing); and (2) that a definitional instruction on “reasonable doubt” is required in light of our decision to abandon the analytical construct.
The State argues that the use of the “reasonable hypothesis analytical construct” conflicts directly with the command of Jackson v. Virginia, 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979); that it distorts the meaning of “proof beyond a reasonable doubt”; and that it continues to imply that circumstantial and direct evidence are inherently different.
The appellant argues that the conscious choice of this Court to use the “reasonable hypothesis analytical construct” is a reasonable method of achieving the realization of Jackson, and is consistent with both the rule of Jackson and the manner in which Texas has decided the issue of sufficiency of the evidence in circumstantial evidence cases for the last 125 years.
In determining that we should bring to an end the requirement that in circumstantial evidence cases, appellate courts must utilize the “reasonable hypothesis analytical construct”, we revisit its history and source.1
*156I. History of the “Analytical Construct”
At least as early as Elizabeth v. State, 27 Tex. 329 (1863), juries in Texas courts have been instructed on the law of circumstantial evidence.2 However, that instruction was repudiated in Hankins v. State, 646 S.W.2d 191 (Tex.Cr.App.1983) (opinion on rehearing). It was explained in Hankins that discontinuance was justified by the reasoning and holding of Holland v. United States, 348 U.S. 121, 75 S.Ct. 127, 99 L.Ed. 150 (1954), a tax evasion case involving the net worth method of proof. Hankins, 646 S.W.2d at 197. In Holland, the Supreme Court held that where the trial court charged the jury on the elements of the offense, the presumption of innocence, the Government’s burden of proof beyond a reasonable doubt (which included a full definition of reasonable doubt), the length and nature of the net worth method of proof and the character of circumstantial evidence in general, it was not necessary for the trial court to add to the charge that “because the proof was circumstantial, the Government must exclude every other reasonable hypothesis than the guilt of the defendant.”
The Hankins Court, citing authorities showing that some twenty-one states and most of the federal circuits had abolished use of the circumstantial evidence charge since the Holland decision, went on to state:
Our holding today finds further support in the fact that there is but one standard of proof for criminal convictions and where the jury is properly instructed on that standard, a charge on circumstantial evidence is valueless and invites confusion. [citations omitted]. The constitutionally required burden of proof of criminal eases “is that the State establish all elements of the offense beyond a reasonable doubt.” Crocker v. State, 573 S.W.2d 190 (Tex.Cr.App.1978) (Opinion on Rehearing). See Mullaney v. Wilbur, 421 U.S. 684, 95 S.Ct. 1881, 44 L.Ed.2d 508 (1975); In re Winship, 397 U.S. 358, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970).... Rather than aiding jurors in applying the reasonable doubt standard, an additional charge on circumstantial evidence focusing on the “reasonable hypothesis” theory serves only to distract jurors from examining the proper standard of proof as the primary focus of their deliberations.
Hankins, 646 S.W.2d at 199.
In 1979, twenty-five years after Holland but four years before Hankins, the Supreme Court decided Jackson v. Virginia, supra, and enunciated a single standard of review for testing the sufficiency of the evidence:
The critical inquiry on review of the sufficiency of the evidence to support a criminal conviction must be not simply to determine whether the jury was properly instructed, but to determine whether the record evidence could reasonably support a finding of guilt beyond a reasonable doubt.... The relevant question is whether after viewing the evidence in the light most favorable to the prosecution, *157any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.
Jackson, 443 U.S. at 318, 99 S.Ct. at 2789.
After Jackson, but again before Han-kins, in Griffin v. State, 614 S.W.2d 155 (Tex.Cr.App.1981), this Court recognized Jackson to be binding on the states as a minimum standard. Speaking for a panel of this Court, Judge Roberts stated:
Although Jackson was setting a standard for review of State convictions by federal courts, the due process requirements that it announced were based expressly on the Fourteenth Amendment. They are binding on the States and constitute a minimum standard for our sustaining a conviction.
Griffin, 614 S.W.2d at 159. And in footnote 5:
Of course States are free to set higher standards of review. For example, this Court requires that juries be instructed not to convict on circumstantial evidence unless it excludes every reasonable hypothesis other than guilt although the federal constitution does not require a different treatment of circumstantial evidence. See Richardson v. State, 600 S.W.2d 818, 826 (Tex.Cr.App.1980) (Dally, J., dissenting, citing Holland v. United States). This different treatment extends to appellate review as well.
Id. at 159.
And in Girard v. State, 631 S.W.2d 162 (Tex.Cr.App.1982), quoting language from Jackson, we stated:
Our task is not to ask whether we believe that the evidence at trial established guilt beyond a reasonable doubt; it is whether, after viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt. Griffin v. State, 614 S.W.2d 155, 159 (Tex.Cr.App.1981). In circumstantial evidence cases, this test is translated into the requirement that the evidence exclude every reasonable hypothesis other than guilt.
See also Allen v. State, 651 S.W.2d 267, 269 (Tex.Cr.App.1983) (finding circumstantial evidence in the case sufficient under Girard and Jackson).3
Griffin, Girard and Allen were decided before the abrogation of the circumstantial evidence charge by the Hankins decision and were predicated upon the fact that, in a circumstantial evidence case, the jury was instructed that they must find that the evidence negated every reasonable hypothesis other than that hypothesis establishing the guilt of the defendant before they could convict. See footnote 1, supra. The jury instruction on circumstantial evidence served as the standard by which they, as factfinder, were to judge the evidence. A reviewing court, in applying Jackson v. Virginia and viewing the evidence in the light most favorable to the jury’s verdict, needed only to determine if the jury acted rationally in finding the evidence sufficient in accordance with the instruction it was given. Once the jury instruction on circumstantial evidence was repudiated in Hankins, however, its function as a guide to juries in reaching their verdict was discontinued.4
*158Shortly after Hankins was decided, this Court handed down Carlsen v. State, 654 S.W.2d 444; Freeman v. State, 654 S.W.2d 450; Denby v. State, 654 S.W.2d 457 and Wilson v. State, 654 S.W.2d 465 (Tex.Cr.App.1983) (opinions on State’s Motions for Rehearing). These four cases, in identical opinions, held that even though it was no longer necessary to instruct juries on circumstantial evidence, the appellate court must still find that every other reasonable hypothesis raised by the evidence was negated, save and except that establishing the guilt of the defendant, if the conviction was to be affirmed in a circumstantial evidence case. The Court, therefore dispensed with the jury charge on circumstantial evidence on the one hand, and kept the “reasonable hypothesis analytical construct” for appellate review on the other. The opinions explained as follows:
It follows that circumstantial evidence should not be tested by an ultimate “standard for review” different from direct evidence; the standard in both kinds of cases is whether “any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt." Accord Griffin v. State, 614 S.W.2d 155 (Tex.Cr.App.1981).
Still, we are unable to devise or discover any reason, compelling or otherwise, for abandoning the utilitarian “exclusion of outstanding reasonable hypotheses” analysis for applying the above “standard for review” in circumstantial evidence cases_ [I]f the evidence supports an inference other than the guilt of the [defendant], a finding of guilt beyond a reasonable doubt is not a rational finding_and this is true irrespective of the character of the evidence.
[W]e are convinced there are no better analytical guidelines for assaying whether a rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt in any given conviction had upon circumstantial evidence than those we currently employ.
Carlsen, 654 S.W.2d at 449-450; Freeman, 654 S.W.2d at 456; Denby, 654 S.W.2d at 464; Wilson, 654 S.W.2d at 471-472 (opinions on rehearing) (emphasis in original).
A majority of this court consistently approved the use of the Carlsen method of reviewing evidentiary sufficiency in circumstantial evidence cases following the Han-kins decision, and recently reiterated in Butler v. State:
It is apparent then, that circumstantial evidence cases have no different standard of review than those cases supported by direct evidence.
[Footnote 1:] We recognize that the United States Supreme Court declined, in Jackson, to adopt [the “reasonable hypothesis”] theory as part of the Jackson standard for review. Likewise, we do not mean to imply an adoption of this theory as the standard of review for sufficiency of the evidence. The reasonable hypothesis theory as utilized by this Court is merely an analytical construct to facilitate the application of the Jackson standard.
769 S.W.2d 234, 238 (Tex.Cr.App.1989) (quoting from Carlsen).
Thus, the state of the law reached somewhat of an analytical quandary at this point since the collective holding of this *159Court was that circumstantial and direct evidence were to have the same standard of review and juries were not to be specially instructed in circumstantial evidence cases, yet appellate courts were to continue using the special aid of the “analytical construct” in the review of circumstantial evidence cases. The classification of the “reasonable hypothesis” legal formula as an “analytical construct” rather than a “higher standard of review” was an attempt at theoretical reconciliation, even though the Court’s earlier decision in Griffin v. State, 614 S.W.2d 155, construed the “reasonable hypothesis” formula to be a higher standard of review than that required by Jackson v. Virginia. Regardless of its label, the fact remains that the abrogation in Hankins of a charge on circumstantial evidence did away with the very basis and authorization for the use of the “reasonable hypothesis” construct in reviewing sufficiency of the evidence in circumstantial evidence cases.5
It is this dichotomy in the law which necessitates our abrogation of the “reasonable hypothesis analytical construct.” As we have emphasized on numerous occasions, the sufficiency of the evidence must be measured against the jury charge. See, e.g., Jones v. State, 815 S.W.2d 667 (Tex.Cr.App.1991) (citing Garrett v. State, 749 S.W.2d 784 (Tex.Cr.App.1986) (opinion on rehearing)). Given the fact that a jury is to be guided by the charge in reaching their verdict, and given the fact that juries are no longer instructed on the law of circumstantial evidence, it no longer makes sense for appellate courts to use the circumstantial evidence “construct” to review the jury’s verdict and to determine, thereby, whether the jurors acted “rationally”. To do so evaluates the jurors’ rationality by a different standard than that by which they were instructed to reach their verdict.
Further, focusing on the existence of an “outstanding reasonable hypothesis inconsistent with the guilt of the accused”, at least where the hypothesis of innocence stems from inconsistencies in the evidence presented at trial, effectively repudiates the jury’s prerogative to weigh the evidence, to judge the credibility of the witnesses, and to choose between conflicting theories of the case. When understood from this perspective, the construct effectively places the reviewing court in the posture of a “thirteenth juror”.6
*160Additionally, use of the construct has led to confusion in the courts of appeals. In the years following Hankins and Carlsen, these courts have grappled with the difficulty of applying the construct and have, on occasion, openly criticized it. See, for example, the majority and dissenting opinions in Ramirez v. State, 801 S.W.2d 110 (Tex.App.—San Antonio 1990) (Majority opinion at 115: “We feel constrained to add that we are uncomfortable with the reasonable hypothesis test”; dissenting opinion at 120: “This court cannot become the ‘thirteenth juror’ and substitute its finding for that of the jury”). See also Moone v. State, 802 S.W.2d 101, 104 n. 3 (Tex.App.—Austin 1990) (“an appellate court’s application of the ‘outstanding reasonable hypothesis’ standard is difficult, at best”).
A prime example of the difficulty and confusion inherent in the application of the analytical construct is that reviewing courts differ widely on the result to be had in applying the construct to the same or a similar set of facts. See, for example, the following cases where this Court disagreed with the court of appeals as to the sufficiency of the circumstantial evidence when considered in light of the analytical construct: Goff v. State, 777 S.W.2d 418 (Tex.Cr.App.1989); Beardsley v. State, 738 S.W.2d 681 (Tex.Cr.App.1987) (plurality); Beier v. State, 687 S.W.2d 2 (Tex.Cr.App.1985); Dickey v. State, 693 S.W.2d 386 (Tex.Cr.App.1984); Burns v. State, 676 S.W.2d 118 (Tex.Cr.App.1984); Jackson v. State, 672 S.W.2d 801 (Tex.Cr.App.1984).7
Another example of confusion is that appellate courts have differed in their eviden-tiary analysis as to the light in which the evidence is to be viewed when reviewing the sufficiency of circumstantial evidence. Some courts have viewed the evidence in the light most favorable to the verdict while others have viewed the evidence in light of the presumption of innocence. See and cf, Carlsen, 654 S.W.2d at 447 (opinion on original submission) and Carlsen, 654 S.W.2d at 449, n.* (opinion on State’s Motions for Rehearing). Although the correct approach was ultimately deemed by this Court to be the former, see id. at 449, there are cases supporting both approaches. See generally Wilson v. State, 654 S.W.2d 465, 468-469 (Tex.Cr.App.1983) (J. Onion’s concurring opinion on original submission); Jackson v. State, 672 S.W.2d 801, 803 (Tex.Cr.App.1984).
Courts have also differed in their interpretation of precisely what evidence is to be considered in reviewing the case for any “outstanding reasonable hypothesis of innocence” — the State’s evidence alone, as suggested in the Carlsen line of cases, or all of the evidence, as suggested by others? See and cf, Carlsen, at 449-450 (“If the State’s evidence supports an inference other than a finding of the essential elements of the crime”), and Houston v. State, 663 S.W.2d 455, 456 (Tex.Cr.App.1984) (“the reviewing court will look at all the evidence in the light most favorable to the verdict”).8 Questions of this nature, as well as questions of general application, have somewhat clouded the jurisprudential history of the analytical construct, and no doubt serve as part of the rationale behind criticisms of the construct as a “confusing” analytical tool.
The analytical construct was rejected by the Supreme Court in Holland, 348 U.S. 121, 140, 75 S.Ct. 127, 137-38, and again in Jackson v. Virginia, 443 U.S. 307, 326, 99 S.Ct. 2781, 2793. It has also been rejected *161as an appellate method of analysis in all federal jurisdictions and many of the State jurisdictions.9 We today reject its use as a method of appellate review for evidentiary sufficiency in this and other appellate courts of this State. We further overrule Carlsen v. State, 654 S.W.2d 444; Freeman v. State, 654 S.W.2d 450; Denby v. State, 654 S.W.2d 457; Wilson v. State, 654 S.W.2d 465; Butler v. State, 769 S.W.2d 234, and their progeny to the extent they conflict with this opinion.
II. Definitional Instruction on “Reasonable Doubt”
Whether the present Court agrees or disagrees with the Hankins decision is not the issue before us. However, in reexamining the two Supreme Court cases which presaged Hankins — the Holland and Jackson decisions — it is evident that both cases implicated the requirement of a full definitional instruction to the jury on reasonable doubt as recognized by Judges Onion, Clinton, and Miller in their individual opinions in the Hankins case. See n. 4, supra at p. 157.10 As Judge Clinton succinctly put it, “the accused is stripped of the benefit of a charge on circumstantial evidence and then loses the protection of a definition on reasonable doubt.” Hankins, 646 S.W.2d at 219.
In his dissent in Hankins, Judge Miller suggested a charge on reasonable doubt in the terms spoken of in Holland, “the kind of doubt that would make a person hesitate to act.” 11 Hankins, 646 S.W.2d at 201, n. *1623 (J. Miller, concurring in part, and dissenting in part). See Holland v. U.S., 348 U.S. at 140, 75 S.Ct. at 137-38. The following definitional instruction on reasonable doubt combines the provisions of Section 2.01, V.T.C.A., Penal Code, and instructions used in the federal system:
All persons are presumed to be innocent and no person may be convicted of an offense unless each element of the offense is proved beyond a reasonable doubt. The fact that a person has been arrested, confined, or indicted for, or otherwise charged with, the offense gives rise to no inference of guilt at his trial. The law does not require a defendant to prove his innocence or produce any evidence at all. The presumption of innocence alone is sufficient to acquit the defendant, unless the jurors are satisfied beyond a reasonable doubt of the defendant’s guilt after careful and impartial consideration of all the evidence in the case.
The prosecution has the burden of proving the defendant guilty and it must do so by proving each and every element of the offense charged beyond a reasonable doubt and if it fails to do so, you must acquit the defendant.
It is not required that the prosecution prove guilt beyond all possible doubt; it is required that the prosecution’s proof excludes all “reasonable doubt” concerning the defendant’s guilt.
A “reasonable doubt” is a doubt based on reason and common sense after a careful and impartial consideration of all the evidence in the case. It is the kind of doubt that would make a reasonable person hesitate to act in the most important of his own affairs.
Proof beyond a reasonable doubt, therefore, must be proof of such a convincing character that you would be willing to rely and act upon it without hesitation in the most important of your own affairs.
In the event you have a reasonable doubt as to the defendant’s guilt after considering all the evidence before you, and these instructions, you will acquit him and say by your verdict “Not guilty”.
See, e.g., 1 E. Devitt & C. Blackmar, Federal Jury Practice and Instructions, § 11.14; Pattern Jury Instructions — 5th Circuit United States Court of Appeals 1988; P. McClung, Jury Charges for Criminal Practice, at p. 6 (1990).
We expressly adopt this instruction on “reasonable doubt” and hold that this instruction shall be submitted to the jury in all criminal cases, even in the absence of an objection or request by the State or the defendant, whether the evidence be circumstantial or direct.12
III. Application to This Case
In this case, the jury was neither instructed on the law of circumstantial evidence nor given a full definitional instruction on reasonable doubt. On appeal of the conviction to the Court of Appeals, that court applied the “analytical construct” to the facts of the case in determining the evidence insufficient to support the verdict. *163Our holding today, abrogating use of the analytical construct, requires that we reverse the Court of Appeals’ judgment. Ordinarily, we would remand this cause to that court for reconsideration in light of our holding that circumstantial cases are to be analyzed solely under the standard of Jackson v. Virginia. However, because the jury was not instructed in accordance with the second part of our holding today, requiring a full definitional instruction on reasonable doubt, we must remand the cause to the trial court for a new trial.
IV. Limited Prospectivity
The Supreme Court, in its recent opinion in Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), decided that “the question ‘of whether a decision [announcing a new rule should] be given prospective or retroactive effect should be faced at the time of [that] decision.’ ” Id. at 489 U.S. 288, 300, 109 S.Ct. 1060, 1069, 103 L.Ed.2d 334, 349 (quoting Mishkin, Foreword: the High Court, the Great Writ, and the Due Process of Time and Law, 79 Harv.L.Rev. 56, 64 (1965)). We agree with this view and therefore address the issue of how our holding is to be applied — prospectively, retroactively, or some variation thereof.13
The Supreme Court, over the past century, has periodically been confronted with claims of retroactivity of new rules pronounced by the Court in its capacity of judicial review. Two types of rules have been the focus of the Court’s retroactivity analyses — rules of constitutional criminal procedure and rules overruling precedent. See generally Annot., Supreme Court Decisions — Retroactivity, 22 L.Ed.2d 821 (1969); Annot., Retroactive or Prospective Overruling, 14 L.Ed.2d 992 (1965). The rules in the instant case are procedural rules overruling precedent. The rules are not of constitutional dimension per se; rather, the rules serve to implement the constitutional requirement that a criminal conviction cannot stand “except upon proof beyond a reasonable doubt.” See In re Winship, 397 U.S. 358, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970). Although the rules obviously address the constitutionally required burden of proof in a criminal case, they do not change it. Neither rule increases or diminishes the requirement of In re Winship that a conviction cannot stand except upon proof of guilt beyond a reasonable doubt.
In Linkletter v. Walker, 381 U.S. 618, 85 S.Ct. 1731, 14 L.Ed.2d 601 (1965), the Supreme Court fashioned a three-pronged test for determining whether a new rule applies retroactively:
(1) the purpose of the rule;
(2) the reliance placed upon the previous doctrine; and
(3) the effect on the administration of justice of a retrospective application of the new rule.
Id. See also Stovall v. Denno, 388 U.S. 293, 297, 87 S.Ct. 1967, 1970, 18 L.Ed.2d 1199 (1967). In various circumstances, the Court applied the three-pronged test and denied full retroactivity to a “new rule.” 14
*164This general approach is sometimes referred to as the “doctrine of nonretroactivity”. One variation of nonretroactivity is to apply the new rule to the parties in the case in which the rule is pronounced and all future cases, but not those pending on direct review or not yet final at the time of the decision, and not to those seeking collateral review of convictions which were final before the new rule was announced. We refer to this approach as “limited pros-pectivity.” See n. 9, supra, at p. 161.
Most recently, however, in Griffith v. Kentucky, 479 U.S. 314, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987), the Court departed from its “doctrine of nonretroactivity” and applied the new rule of Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. 1712, 90 L.Ed.2d 69 (1986), retrospectively to all cases pending on direct federal and state review or not yet final at the time Batson was decided. In Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), the Supreme Court held that Batson does not apply retroactively on collateral review of convictions that became final before the Batson opinion was announced. See also Allen v. Hardy, 478 U.S. 255, 106 S.Ct. 2878, 92 L.Ed.2d 199 (1986).
These holdings respecting the retroactivity of the Batson rule adopt the twofold approach suggested by Justice Harlan in his dissent in Desist v. United States, 394 U.S. 244, 256, 89 S.Ct. 1030, 1037, 22 L.Ed.2d 248, 259 (1969), and represent the current viewpoint of the Supreme Court respecting retroactivity — a new rule will apply to all future cases, the case announcing the rule, and all cases then pending on direct review or not yet final, but not to the collateral review of cases which were final before the new rule was announced. Under this view, habeas petitions should generally be judged according to the law in effect at the time of the conviction; all other cases are decided in accordance with the new rule. We refer to this approach as “limited retroactivity.” See n. 9, supra, at p. 161.
That said, however, we must note that the posture of the Supreme Court respecting the retroactivity of judicially-announced rules is neither binding on this Court nor mandated by the federal constitution. As has been frequently recognized by the Supreme Court, “the Constitution neither prohibits nor requires retrospective effect [of a new rule] ... ‘[T]he federal constitution has no voice upon the subject’.” Linkletter v. Walker, 381 U.S. 618, 629, 85 S.Ct. 1731, 1737-38, 14 L.Ed.2d 601, 608 (1965).15 The decision whether to apply a new rule retrospectively “is not a matter of constitutional compulsion, but a matter of judicial policy, to be determined by the court after weighing the merits and demerits of the particular case, by looking to the prior history of the rule in question, its purpose and effect, and whether retroactive application will further or retard its operation.” 16 Annotation, Retroactive or Prospective Overruling, 14 L.Ed.2d 992 (1965) (paraphrasing Linkletter, 381 U.S. at 629, 85 S.Ct. at 1737-38, 14 L.Ed.2d at 608).
Applying these basic principles to the new rules at issue compels the conclu*165sion that justice is best served by an application of “limited prospectivity,” i.e., application of the rules will be limited to the case at bar and all cases tried hereafter. We want to emphasize that we are not adopting this approach for all cases which establish new rules. In general, we agree with the Supreme Court’s rationale in Griffith where the Court adopted the “limited retrospectivity” approach, to use our nomenclature. See Griffith v. Kentucky, 479 U.S. 314, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987). Ketroactivity, of course, must be decided on a case-by-case basis.
For the following reasons, we limit our use of the “limited prospectivity” approach to the case at bar. First, the rules at issue in the instant case are distinguishable from those in Griffith. The Griffith rationale was keyed to the “actual inequity” that results when only one of many similarly situated defendants receives the benefit of a new rule. Griffith, 107 S.Ct. at 714 (quoting United States v. Johnson, 457 U.S. 537, at 555, n. 16, 102 S.Ct. 2579, at 2590, n. 16, 73 L.Ed.2d 202 (1982)). The rule of this case, however, of adhering to the minimal standard of review enunciated in Jackson v. Virginia, was instituted on petition by the State, not at the defendant’s request.
The corresponding institution of the new rule requiring that trial courts provide juries in criminal cases with a full definitional instruction on “reasonable doubt” is simply a necessary corollary to the Jackson standard of review, given that the sufficiency of the evidence is measured in great part by the court’s instructions to the jury and the degree of rationality and reasonableness utilized by the jury in applying those instructions to the evidence. However, the requirement of an instruction on “reasonable doubt” cannot be said to be a greater benefit to the defendant than was the “reasonable hypothesis of innocence analytical construct.” Thus, unlike the issue in circumstantial evidence cases presented in Griffith, no “actual inequity” results from denying a retrospective application of these new rules.
Secondly, the purpose of the rules in this case — more logically correlate the standard of review for sufficiency claims with the instructions provided in the court’s charge to the jury — is best served by a prospective application. The rules, being procedural in nature and not conferring any greater constitutional protections than existed before, are not of the type which typically militate in favor of a retrospective application. The mechanics of implementing the rules militate against a retrospective application because of the reliance placed upon the previous doctrine of analyzing the sufficiency of circumstantial cases under the analytical construct. In the various cases already pending on direct review or not yet final, juries were not instructed with a full definition of reasonable doubt. Thus, the purpose of correlating the standard of review with the court’s instructions to the jury cannot be effected retroactively without requiring new trials in each of the affected cases.
Lastly, this type of retrospective application would unduly burden the administration of justice. It makes more sense, in the unique context of this case, to allow defendants in the cases pending on direct review or not yet final to have their cases decided under the analytical construct than to require new trials in every case for failure to charge the jury with a full definitional instruction on reasonable doubt. The latter approach would not further the purpose of the new rules and would unnecessarily hinder the administration of justice.
For these reasons, we adopt the new rules and apply them to the instant case and all cases tried hereafter. Accordingly, the judgment of the Court of Appeals is reversed and the cause is remanded to the trial court for a new trial.
. For a more thorough review of the Texas history of the law on circumstantial evidence and its treatment both as a standard of appellate review and as a cautionary jury instruction, see Judge Onion’s dissenting opinion in Hankins v. State, 646 S.W.2d 191, 203 (Tex.Cr.App.1983).
. The typical circumstantial evidence charge in Texas read as follows:
This is a case depending for conviction on circumstantial evidence. In order to warrant a conviction of a crime on circumstantial evidence, each fact necessary to the conclusion sought to be established must be proved by competent evidence beyond a reasonable doubt; all the facts (that is, the facts necessary to the conclusion) must be consistent with each other and with the main fact sought to be proved, and the circumstances, taken together, must be of a conclusive nature, leading on the whole to a satisfactory conclusion and producing, in effect, a reasonable and moral certainty that the accused, and no other person, committed the offense charged.
But in such cases it is not sufficient that the circumstances coincide with, account for and therefore render probable, the guilt of the defendant. They must exclude, to a moral certainty, every other reasonable hypothesis except the defendant’s guilt, and unless they do so, beyond a reasonable doubt, you will find the defendant not guilty.
See, e.g., Hunter v. State, 142 Tex.Crim. 224, 152 S.W.2d 365 (1941); Harroll v. State, 135 Tex.Crim. 65, 117 S.W.2d 103 (1938); Kunde v. State, 22 Tex.App. 65, 3 S.W. 325 (1886); Rogue v. State, 12 Tex.App. 283 (1882); Hampton v. State, 1 Tex.App. 652 (1877); Perkins v. State, 32 Tex. 110 (1869).
. Thus, in Texas at least, the "reasonable hypothesis analytical construct" was envisioned as an effective means of implementing the Jackson standard of review in circumstantial evidence cases. This departure from our federal counterparts was well within the state’s prerogative, of course, as Jackson only propounded a minimum standard of review and Holland was not constitutionally binding on the states. For a listing of the states which, at the time of the Hankins decision, had chosen to follow Holland in abolishing the requirement of a charge on circumstantial evidence "where the jury is properly instructed on the reasonable doubt standard," see Hankins, 646 S.W.2d at 197.
. The Hankins decision was not without criticism. Judge Onion, in his dissenting opinion, stated that the majority based its renouncement of the circumstantial evidence charge on the fact that "the giving of the long approved circumstantial evidence charge is now somehow inherently confusing even when considered with a non-definitional charge on reasonable doubt". Hankins, 646 S.W.2d at 204. Without agreeing with the majority’s conclusion, he pointed out that the presumption underlying the Holland doctrine is that an adequate instruction on the standard for reasonable doubt fulfills the function of the cautionary instruction on the law of circumstantial evidence and possibly re-*158suits in less jury confusion in evaluating direct and circumstantial evidence. Id. at 210.
Judge Miller, concurring in the abolition of the circumstantial evidence charge but dissenting because the Court did not also require a full definitional charge on reasonable doubt, stated:
I do not agree with the majority’s implied assumption that a jury is not confused by the naked and undefined term "reasonable doubt.” In other words, the majority does not go far enough to clear up the confusion of which they complain.
Hankins, 646 S.W.2d at 201. Likewise, Judge Marvin O. Teague, in his prophetic dissent in Hankins, stated:
Today’s opinion will cause confusion to exist in the appellate courts of this State as the majority has left open the question of what standard of review this Court will apply in future circumstantial evidence cases ... [I]f this court continues to apply the former standard of review, when a jury in a particular case was not instructed to apply that standard, I foresee a sharp increase of needless reversals by the appellate courts of this State.
Hankins, 646 S.W.2d at 220.
. The dissent challenges this rationale on the basis that "in this jurisdiction a standard of appellate review for evidentiary sufficiency in a circumstantial evidence case preceded any requirement that a jury be thus charged," dissenting opinion at p. 173, n. 13 (emphasis in original), and cites to excerpted language from the Texas decision of Henderson v. State, 14 Tex. 503 (1855), quoting language from the case of Commonwealth v. Webster, 59 Mass. (5 Cushing) 295, 52 American Decisions 711 (1850), generally considered to be the source of the cautionary jury instruction on the law of circumstantial evidence. See generally, Hankins v. State, 646 S.W.2d 191, 20'5 (Tex.Cr.App. 1981) (J. Onion’s dissenting opinion).
We must point out, however, that the Webster case is not a reported legal opinion reviewing the verdict in that case, but is in fact a verbatim report of the actual charge given to the jury in that case by Chief Justice Shaw. As Judge Onion explained in his dissenting opinion in Han-kins, 646 S.W.2d at 205, ”[i]n Henderson the court quoted with approval the test from Webster, which was and is the basis for the cautionary jury instruction on the law of circumstantial evidence.” The Henderson case relies on the Webster charge in adopting the rule for reviewing the sufficiency of circumstantial evidence. The necessity of the charge itself was not raised in Henderson. See Henderson, 14 Tex. at 506 and 520. The fact remains that the jury instruction and the standard of review are logically correlative and can be fairly assessed as interdependent.
. Indeed, at least one court has misinterpreted the construct as requiring that "[a] conviction will not be sustained on appeal if ... the evidence does not produce in the mind of the appellate court a firm conviction or ‘moral certainty’ of the guilt of the [defendant].” Jackson v. State, 672 S.W.2d 801, 803 (Tex.Cr.App.1984) (quoting opinion by court of appeals). In reversing the judgment of the court of appeals, we explained that the Jackson v. Virginia standard of review applies to both circumstantial and direct cases and requires the reviewing court to determine whether “any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt,” not whether the reviewing court could so find. Id. at 803. Although erroneous, the court's reasoning is understandable in light of the tension at play between the jury’s prerogative to disbelieve a reasonable hypothesis of innocence raised by circumstantial evidence, and the reviewing court’s duty to reverse where that reasonable hypothesis is extant.
. This difficulty was highlighted in the cases of Markham and Phelps, two companion cases involving co-defendants convicted on the basis of circumstantial evidence, where a panel of the Court of Appeals reversed the conviction in one case, holding the evidence insufficient utilizing the analytical construct, while another panel of the Court of Appeals upheld the conviction in the companion case, as constituting sufficient evidence utilizing the analytical construct. See Markham v. State, 761 S.W.2d 553 (Tex.App.—San Antonio 1988), and cf. Phelps v. State, 730 S.W.2d 198 (Tex.App.—San Antonio 1987) (no. pet.).
. We note that in Jackson v. Virginia, the Supreme Court made clear that a reviewing court is to consider "all of the evidence " in assessing its sufficiency to support the verdict. Jackson v. Virginia, 443 U.S. at 319, 99 S.Ct. at 2789, 61 L.Ed.2d at 573 (“upon judicial review all of the evidence is to be considered in the light most favorable to the prosecution”) (emphasis in original).
.Federal jurisdictions: U.S. v. Bell, 678 F.2d 547 (5th Cir.1982); U.S. v. Lechuga, 888 F.2d 1472 (5th Cir.1989); U.S. v. McNatt, 813 F.2d 499 (1st Cir.1987); U.S. v. Elsbery, 602 F.2d 1054 (2nd Cir.1979); U.S. v. Hamilton, 457 F.2d 95 (3rd Cir.1972); U.S. v. Chappell, 353 F.2d 83 (4th Cir.1965); U.S. v. Conti, 339 F.2d 10 (6th Cir.1964); U.S. v. Wigoda, 521 F.2d 1221 (7th Cir.1975); U.S. v. Carlson, 547 F.2d 1346 (8th Cir.1976); U.S. v. Nelson, 419 F.2d 1237 (9th Cir.1969); U.S. v. Merrick, 464 F.2d 1087 (10th Cir.1972); U.S. v. Poole, 878 F.2d 1389 (11th Cir.1989); U.S. v. Davis, 562 F.2d 681 (D.C.Cir.1977).
State jurisdictions: Des Jardins v. State, 551 P.2d 181 (Alaska 1976); State v. Nash, 143 Ariz. 392, 694 P.2d 222 (1985); Ciccaglione v. State, 474 A.2d 126 (Del.1984); Ford v. U.S., 498 A.2d 1135 (D.C.Ct.App.1985); Youngblood v. State, 179 Ga.App. 163, 345 S.E.2d 634 (1986); People v. Eyler, 133 Ill.2d 173, 139 Ill.Dec. 756, 549 N.E.2d 268 (1989); Kidd v. State, 530 N.E.2d 287 (Ind.1988); State v. Radeke, 444 N.W.2d 476 (Iowa 1989); State v. Morton, 230 Kan. 525, 638 P.2d 928 (1982); Hines v. State, 58 Md.App. 637, 473 A.2d 1335 (1984); People v. Johnson, 137 Mich.App. 295, 357 N.W.2d 675 (1984); Stokes v. State, 518 So.2d 1224 (Miss.1988); State v. Buchanan, 210 Neb. 20, 312 N.W.2d 684 (1981); State v. Sutphin, 107 N.M. 126, 753 P.2d 1314 (1988); State v. Jones, 303 N.C. 500, 279 S.E.2d 835 (1981); State v. Jacobson, 419 N.W.2d 899 (N.D.1988); State v. Jenks, 61 Ohio St.3d 259, 574 N.E.2d 492 (1991); State v. Duran, 107 N.M. 603, 762 P.2d 890 (1988); Commonwealth v. Sullivan, 472 Pa. 129, 371 A.2d 468, 478 (1977); State v. Caruolo, 524 A.2d 575 (R.I.1987); State v. Neale, 145 Vt. 423, 491 A.2d 1025 (1985); State v. Couch, 44 Wash.App. 26, 720 P.2d 1387 (1986); State v. Poellinger, 153 Wis.2d 493, 451 N.W.2d 752 (1990).
. Until today, "reasonable doubt" has never been defined in Texas, either statutorily or in the charge to the jury. See V.T.C.A., Penal Code, Section 2.01. After the enactment of Section 2.01 of the Texas Penal Code in 1974, the following “nondefinitional” charge on presumption of innocence and reasonable doubt was given:
All persons are presumed to be innocent and no person may be convicted of an offense unless each element of the offense is proved beyond a reasonable doubt. The fact that he has been arrested, confined or indicted for, or otherwise charged with the offense, gives rise to no inference of guilt at this trial. In case you have a reasonable doubt as to the defendant’s guilt after considering all the evidence before you, and these instructions, you will acquit him and say by your verdict "not guilty.”
See Carr v. State, 600 S.W.2d 816 (Tex.Cr.App.1980).
. As Judge Miller pointed out in Hankins, the definition of "reasonable doubt" is clarified when one compares it with the lesser burdens of proof:
This definition of reasonable doubt ... becomes almost crystal clear when put in perspective with the burdens of proof of “preponderance of the evidence" and of "clear and convincing evidence.” ...:
Proof by a preponderance of the evidence ... is defined as that degree of proof that, when taken as a whole, shows that a fact sought to be proved is more probable than not.
Proof by clear and convincing evidence ... is defined as that degree of proof which will produce in the jury*s mind a firm belief as to the truth of the allegation sought to be established. This is an intermediate standard, fall*162ing between the preponderance of the evidence standard and the reasonable doubt standard, [citation omitted].
Hankins, 646 S.W.2d at 203, n. 6.
. In framing a definitional instruction on reasonable doubt, we are cognizant of the recent decision of the Supreme Court in Cage v. Louisiana, — U.S. —, 111 S.Ct. 328, 112 L.Ed.2d 339 (1990), where the Court reversed the conviction because the reasonable doubt instruction was contrary to In re Winship, 397 U.S. 358, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970), which requires that a conviction must be based on proof beyond a reasonable doubt of every fact necessary to constitute the crime with which the defendant is charged.
In Cage, the Supreme Court emphasized that the Louisiana charge equated reasonable doubt with “grave uncertainty " and “actual substantial doubt” and, emphasizing the words "substantial" and "grave,” said that they suggested a higher degree of doubt than is required for acquittal under the reasonable doubt standard. Cage, 111 S.Ct. 328. The Supreme Court concluded that it was clear a “reasonable” juror could have interpreted the instruction to allow a finding of guilt based on a degree of proof below that required by Winship. Id. The definitions we adopt today are not prohibited by the Supreme Court's decision in Cage.
. For the purposes of this opinion, the following definitions will apply: (1) pure prospectivity — applying the new rule only to those cases tried after the effective date of the opinion announcing the new rule; (2) limited prospectivity — applying the new rule to the parties in the case in which the new rule is announced and to those cases tried after the effective date of the opinion announcing the new rule; (3) limited retrospectivity — applying the new rule to all cases then pending on direct review or not yet final, as well as those in (2); and (4) pure retrospectivity — applying the new rule to cases pending on collateral review as well as those included within (3).
It should be noted that the nomenclature employed to describe the different approaches varies among the courts and authorities. The label of "retroactive”, for example, has sometimes been used to describe an application of a new rule to the parties in the case announcing the rule. See generally Annot., Supreme Court Decisions — Retroactivity, 22 L.Ed.2d 821 (1969) (collecting and analyzing United States Supreme Court cases in which the court has expressed its views as to the retroactive effect of its own decisions announcing new rules).
. The Court has "drawn the retroactivity line in a variety of places," as detailed in Solem v. Stumes, 465 U.S. 638, 650-651, 104 S.Ct. 1338, 1345-46, 79 L.Ed.2d 579 (1984):
Some decisions have been applied only to defendants whose convictions were not yet final when the new rule was established, United States v. Johnson, 457 U.S. 537, 73 L.Ed.2d *164202, 102 S.Ct. 2579 (1982); Linkletter v. Walker, 381 U.S. 618, 14 L.Ed.2d 601, 85 S.Ct. 1731 (1965), some only to those defendants whose trials had not yet begun at that point, Johnson v. New Jersey, 384 U.S. 719, 16 L.Ed.2d 882, 86 S.Ct. 1772 (1966); DeStefano v. Woods, 392 U.S. 631, 20 L.Ed.2d 1308, 88 S.Ct. 2093 (1968), some only to those whose constitutional rights were violated after the law-changing decision was handed down, United States v. Peltier, 422 U.S. 531, 45 L.Ed.2d 374, 95 S.Ct. 2313 (1975); Desist v. United States, 394 U.S. 244, 22 L.Ed.2d 248, 89 S.Ct. 1030 (1969); Stovall v. Denno, 388 U.S. 293, 18 L.Ed.2d 1199, 87 S.Ct. 1967 (1967), and some only to those cases where the prosecution sought to introduce (newly) illegal evidence after the date of the nonretroactive decision, Fuller v. Alaska, 393 U.S. 80, 21 L.Ed.2d 212, 89 S.Ct. 61 (1968).
. Quoting Justice Cardozo in Great Northern R. Co. v. Sunburst Oil & Refining Co., 287 U.S. 358, 364, 53 S.Ct. 145, 148, 77 L.Ed. 360, 366 (1932) (referring to state court’s prospective overruling of prior decision).
. The Supreme Court’s subsequent decisions in Griffith and Teague did not alter or overrule this earlier language in Linkletter regarding the fact that retroactivity is not a constitutional issue, but one of judicial policy.