The opinion of the Court was delivered by CLIFFORD, J.

We granted certification, 63 N.J. 501 (1973), to review the Appellate Division's affirmance of the county court's order of incarceration on reinstatement of a six month jail sentence for defendant's violation of probation. Placed squarely in issue is the applicability of this Court's decision in State v. DeBonis, 58 N.J. 182 (1971), which held that a defendant who appeals from a municipal court conviction may not receive a greater sentence in the county court. It is clear that prior to DeBonis a trial de novo in the county court involved not only the issue of guilt or innocence but also the exercise of the independent judgment of that court in the matter of sentence. State v. States, 44 N.J. 285, 293 (1965). We hold that the DeBonis rule does apply under the facts of this case and to all cases on direct appeal as of the date that decision was announced. This leads to a reversal of the judgment below.

In 1969 defendant, Herman Nash, was a teacher in Teaneck High School. His difficulties with the law arise out of an occurrence of March 6 of that year when he led a group of about 15 of his students into the principal's office in protest against a school policy under which lower-track, general science students were denied use of laboratory facilities. For more than eight years defendant had tried unsuccessfully to change this policy, but he had never been able to obtain regularly scheduled laboratories for this group of students, despite the fact that several such labs were scheduled each week for the better students.

His frustration led to the demonstration. No violence was involved but some disorder was created. The participants cleaned up the principal's office before leaving. However, the "sit-in" led to Nash's conviction in the Teaneck Municipal Court for the violation of N.J.S.A. 2A:170-28 (disturbing a place of assembly), N.J.S.A. 2A:170-31 (trespass), and N.J.S.A. 2A:170-26 (assault). He was fined $250 each on the first two charges and sentenced to a thirty day jail term on the assault charge. These convictions were appealed to the county court, which conducted a plenary trial de novo by order of the assignment judge on defendant's application.

The county court dismissed the assault and trespass charges but reconvicted defendant under N.J.S.A. 2A:170-28 for disturbing a place of assembly. However, instead of receiving a fine as originally imposed, he was sentenced on December 12, 1969 to a six-months jail term, suspended in favor of a three-year probationary period. An appeal taken from this conviction was argued in the Appellate Division on January 11, 1971. On April 8, 1971, this Court's decision in State v. DeBonis, supra, was handed down. On May 12, 1971 the Appellate Division affirmed defendant's conviction and certification was denied by this Court on July 7, 1971, 58 N.J. 597.

For reasons not explained in the record before us Nash was not called to report to the probation office until March 24, 1972. Although he did appear punctually, he refused "as a matter of principle" to sign the rules and regulations, as required by R. 3:21-7. However, he agreed to abide by the conditions of his probation, and asserted, without contradiction, that he had in fact been abiding by them since the time of his sentence in December 1969. The only special condition which had been imposed upon him was that he not attend board of education meetings in Teaneck during his probationary period. After a hearing on April 28, 1972 on a motion to modify the sentence, the county court revoked defendant's probation and re-imposed the jail term, solely as a consequence of his refusal to sign the probation department rules.

Thereafter, on May 5, 1972 the defendant appealed the probation revocation to the Appellate Division, petitioning also for a stay of his sentence, which was granted. On May 22, 1972 he petitioned the county court for post-conviction relief, based on DeBonis. That petition was subsequently dismissed. The Appellate Division, on July 13, 1972, granted a motion to consolidate an appeal from the denial of the post-conviction relief with the appeal from the probation revocation, in order, as it said, "to dispose of all issues once and for all." As indicated hereinabove, that court, in an unreported opinion handed down on April 10, 1973, affirmed both the conviction and sentence of the county court and also the revocation of probation. Our conclusion that DeBonis applies to this defendant results in the vacating of the jail sentence.

While the Appellate Division speaks in terms of the non-retroactivity of DeBonis, it is clear that the decision below is simply a determination that the DeBonis rule should not be applied in the particular circumstances of this case, not a declaration of its non-retroactivity as a matter of legal principle or policy. That Court reasoned that since Nash "had an ample opportunity effectively to raise the DeBonis ruling in his direct appeal from the conviction and sentence and in his application for certification which followed, he should not be again permitted an additional opportunity to argue the matter * * *."

The issue of the retroactivity of DeBonis clearly intrudes itself into this case, just as it did in State v. Koch, 118 N.J. Super. 421 (App.Div. 1972), final judgment entered in 119 N.J. Super. 184 (App.Div. 1972). The court there was faced with the problem of retroactivity of Burgett v. Texas, 389 U.S. 109, 88 S.Ct. 258, 19 L.Ed.2d 319 (1967) (non-activation of a repetitive-offender type statute to enhance punishment where the prior convictions were uncounselled and there was no waiver of counsel). The Koch court observed that the retroactivity question has two facets, which are somewhat interrelated:

"The first is whether defendant could invoke Burgett in the present post-conviction proceedings, within the intent of R. 3:22 regulating post-conviction relief. The second is whether general principles of retroactivity of newly adjudicated federal constitutional rights are compatible with the allowance of the instant attack." 118 N.J. Super. at 429.

We deal with those "two facets" in the reverse order, addressing ourselves first to general principles of retroactivity and their applicability here, and secondly to the availability of the DeBonis rule in these proceedings.

I

Four common approaches to retroactivity emerge from the case law, and no distinction is apparent simply on the basis that civil or criminal litigation is involved. See Linkletter v. Walker, 381 U.S. 618, 627, 85 S.Ct. 1731, 1736, 14 L.Ed.2d 601, 607 (1965). A variety of approaches has been taken to the question of retroactive effect of court decisions in the criminal field recognizing or creating new rights of persons investigated for, charged with or prosecuted for crime. These have depended largely on the court's view of what is just and consonant with public policy in the particular situation presented. The four groupings of rules which have evolved are as follows:

1. Some modern decisions recognize the court's power to hold that an overruling decision operates prospectively only and is not even operative upon the rights of the parties to the case declaring the new rule of law. E.g., James v. United States, 366 U.S. 213, 81 S.Ct. 1052, 6 L.Ed.2d 246 (1961) (embezzled funds constitute taxable income); England v. Louisiana State Board of Medical Examiners, 375 U.S. 411, 84 S.Ct. 461, 11 L.Ed.2d 440 (1964) (under abstention doctrine litigant must preserve right to return to Federal court).

2. There are cases which limit the retroactive effect of a new rule, so that it will govern the rights of the parties to the overruling case itself but apply the overruled law to the rights of parties to all other pending litigation as well as litigation terminated by final judgment. E.g., Stovall v. Denno, 388 U.S. 293, 87 S.Ct. 1967, 18 L.Ed.2d 1199 (1967) (right to counsel at out-of-court identification proceedings); Desist v. United States, 394 U.S. 244, 89 S.Ct. 1030, 22 L.Ed.2d 248 (1969) (exclusionary rule for electronic surveillance without court order); Michigan v. Payne, 412 U.S. 47, 93 S.Ct. 1966, 36 L.Ed.2d 736 (1973) (reasons must be indicated for greater sentence in second trial after reversal).

3. Another form of limited retroactive effect permits the new rule to govern the rights of the parties to the overruling case and the rights of parties to other cases pending, in the sense of final direct review not having been exhausted, when the overruling case was decided. The old rule applies to the rights of parties in cases terminated by a final judgment no longer appealable at the time of the decision in the case declaring new law. E.g., Linkletter v. Walker, supra, (exclusion of illegally seized evidence); Tehan v. Shott, 382 U.S. 406, 86 S.Ct. 459, 15 L.Ed.2d 453 (1966) (prosecutor's comment on defendant's failure to testify); State v. Koch, supra, (use of uncounselled prior conviction to affect credibility).

4. Finally, a new rule may be given general retroactive effect so as to apply even where final judgments have been obtained as of the time the overruling case is decided. E.g., Eskridge v. Washington State Board, 357 U.S. 214, 78 S.Ct. 1061, 2 L.Ed. 2d 1269 (1958) (free transcript provided to indigents appealing convictions); Pickelsimer v. Wainwright, 375 U.S. 2, 84 S.Ct. 80, 11 L.Ed.2d 41 (1963) (right to counsel at trial).

As the first paragraph of this opinion indicates, we adopt in this case, for the reasons expressed here below, the third approach, holding DeBonis retroactive to all cases pending on direct appeal when that decision was announced on April 8, 1971.

The question of which approach to retroactivity will be taken in a given case is not answered by any constitutional mandate. E.g., Linkletter v. Walker, supra; State v. Smith, 37 N.J. 481 (1962), cert. den. 374 U.S. 835, 82 S.Ct. 1879, 10 L.Ed.2d 1055 (1963). Rather, under the view adopted by the United States Supreme Court, the competing considerations in each case are weighed by examining (1) the purpose of the rule and whether it would be furthered by a retroactive application, (2) the degree of reliance placed on the old rule by those who administered it, and (3) the effect a retroactive application would have on the administration of justice. This weighing process has generally been followed in New Jersey; see, e.g., State v. Johnson, 43 N.J. 572 (1965), aff'd, 384 U.S. 719, 86 S.Ct. 1772, 16 L.Ed.2d 882 (1966); Darrow v. Hanover Twp., 58 N.J. 410 (1971); State v. Koch, supra.

Of particular relevance to any weighing of these competing considerations in the present context are two recent United States Supreme Court cases, North Carolina v. Pearce, 395 U.S. 711, 89 S.Ct. 2072, 23 L.Ed.2d 656 (1969), and Michigan v. Payne, supra. Pearce held that a sentence could not be increased at the second trial where a judge was doing the re-sentencing and where he failed to state his reasons for the increased sentence on the record. The holding was necessary, the court said, to prevent vindictiveness from playing any part in the re-sentencing determination.

Michigan v. Payne, supra, held that Pearce was not retroactive. The Court reasoned that Pearce was like Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966) in laying down a prophylactic rule, here to assure that sentencing is based only on proper considerations. Thus, if Pearce were applied retroactively, the result would be an unjustifiable benefit to prisoners in whose second sentence vindictiveness had played no part. The opinion emphasized that a re-sentenced prisoner could establish his eligibility for post-conviction relief by demonstrating that vindictiveness did in fact play a part in his sentencing. However, prisoners re-sentenced before Pearce would not be eligible for automatic remand for resentencing. Additionally, judges reasonably relied on pre- Pearce law and had no reason to state reasons for levying a higher sentence after a second trial. Therefore, the Payne court felt that the effect of a retroactive application of Pearce on the administration of justice would be substantial. It would require sentencing judges to make factual determinations on sentences meted out years before, a task which often would be impossible.

However, we do not believe that the Supreme Court's reasoning in Payne is dispositive of the issue of the retroactivity of DeBonis. Both DeBonis and Pearce deal with questions of legality of increased sentences. There the similarity ends. Pearce was a constitutionally-based decision, required, the Court said, by due process. DeBonis is a decision grounded on the strong state policy that those tried in municipal courts should not suffer possible injustices arising from that admittedly imperfect judicial system. Rather, they should be free to appeal municipal court convictions to the county courts. Thus, the purpose of the DeBonis decision is entirely different from the purpose of the Pearce decision. DeBonis is additionally different from Pearce in the effect its retroactive application would have on the administration of justice. Retroactive application of Pearce would mean that all judges who had imposed harsher sentences the second time around would either have to file a statement of their reasons or have the sentences vacated. This would pose a hardship as a general rule because of the great number of sentencing hearings which would be necessary, and would create a special burden in cases where the sentencing had taken place long before. However, in the DeBonis situation no new hearings would have to be held since the first sentence could be applied automatically. Additionally, no judge would have to issue a statement of reasons for the harsher sentence since all harsher sentences are automatically illegal under DeBonis. Moreover, the number of cases which would be affected by a retroactive application would be minimal since criminal actions over which municipal courts exercise jurisdiction are less serious than those brought in county courts and custody sentences are likely to be short. A retroactive application of Pearce would have extended to many more cases.

Finally, county court judges had good reason not to rely too heavily on pre- DeBonis law, since it was presaged both by United States Supreme Court decisions and by holdings of this Court. Pearce itself should have given grounds for doubt, as should this Court's decision in State v. Wolf, 46 N.J. 301 (1966), which relied on New Jersey policy to declare that on retrial a defendant could not be sentenced to death if he had not been so sentenced after his first trial. Payne, on the other hand, declares that Pearce's holding was not foreshadowed by any previous decision of the Supreme Court. 412 U.S. at 55, 93 S.Ct. at 1971, 36 L.Ed.2d at 745.

Therefore, a weighing of the competing considerations ante leads us to conclude that because of New Jersey's strong policy, described above, against making a criminal defendant's cause depend on the "antiquated system of local courts," State v. DeBonis, supra, 58 N.J. at 188-189; because this policy can only be strengthened, not weakened, by giving DeBonis limited retroactive application; and because there are no countervailing considerations, such as an undue administrative burden which would result from such a decision, we should employ the rule of limited retroactivity referred to hereinabove.

II

We turn now to the second facet of retroactivity, i.e., whether defendant can raise the retroactivity issue at this time in a post-conviction proceeding, after his direct appeal has been exhausted. The State argues that because DeBonis was handed down on April 8, 1971, Nash had ample time to raise the issue before certification was denied by this Court on July 7, 1971. As pointed out heretofore, this view was concurred in by the Appellate Division, which affirmed the denial of post-conviction relief on this narrow basis. In our view this position misapplies R. 3:22-4, reading as follows:

Any ground for relief not raised in a prior proceeding under this rule, or in the proceedings resulting in the conviction, or in a post-conviction proceeding brought and decided prior to the adoption of this rule, or in any appeal taken in any such proceedings is barred from assertion in a proceeding under this rule unless the court on the motion or at the [final] hearing finds (a) that the ground for relief not previously asserted could not reasonably have been raised in any prior proceeding (b) * * *.

Defendant's attorneys go to great lengths to explain why they were not aware of DeBonis at an earlier point in time, when they could have made use of it on his direct appeal. They suggested at oral argument that among other things they were not alerted, while the direct appeal was in progress, to the threat of incarceration, inasmuch as the six-months jail sentence had been suspended. It was not until after defendant refused to sign the rules and regulations that imprisonment became a very real threat. While we do not look favorably upon their unawareness of a newly-announced New Jersey decision, we do not believe that the period between the time that DeBonis was handed down and the time when Nash's direct appeal was concluded was so long as to make them inexcusably remiss in their professional duties in failing to cite it. Additionally, we hesitate to make the availability of a retroactive principle in a criminal context turn on whether an attorney has read recent advance sheets. Thus, we permit the defendant in this case to have the benefit of the DeBonis ground for relief by holding that under the circumstances herein that ground "could not reasonably have been raised in any prior proceeding," R. 3:22-4. But cf. State v. Koch, supra.

The judgment of the Appellate Division is reversed. Defendant's sentence is vacated and the case remanded to the Bergen County Court for resentencing in accordance with this opinion.

For reversal and remandment — Justices JACOBS, SULLIVAN, PASHMAN and CLIFFORD and Judge CONFORD. — 5.

For affirmance — None.