I seek to draw the Commission's attention to a small but important problem with the administration of justice by the Courts of Appeals: the increasing tendency of the appellate courts to render their decisions by unpublished opinions, resulting in a growing body of law which is not accessible to the public.


The genesis of limited publication rules was the 1973 report of the Advisory Council for Appellate Justice(1) urging courts to reduce the number of published opinions. The report recommended the adoption of limited publication plans by the appellate courts, proposing that an opinion be published only if it 1) establishes a new rule of law, or alters or modifies and existing rule, 2) involves a legal issue of continuing public interest, 3) criticizes existing law or 4) resolves a conflict of authority. By 1977, most of the circuits had adopted limited publication plans, although with considerable variation in the criteria used to determine whether a decision should be published, how many votes are required to publish an opinion, and whether unpublished opinions may be cited.(2) Such variations persist more than twenty years after the adoption of limited publication plans.

Limited publication rules have drawn considerable criticism. One commentator suggests that critiques of limited publication plans may be grouped into five broad categories: Judicial Accountability, Review by Higher Court, Predicting Precedential Value, Equality of Access, and Judicial and Litigant Economy.(3) Reynolds and Richman, for example, have attacked limited citation rules as allowing courts to decide behind closed doors "troublesome cases presenting issues the court does not want to address in public". Although critics have demonstrated that there are indeed instances of cases which should have been published but were not, perhaps due to an intentional failure to comply with the rules, none of the critics have undertaken a study of all of the circuits. Thus, they are vulnerable to counter-criticism that there has been no systematic showing of increased "judicial irresponsibility"(4) as a consequence of adoption of limited publication rules.

Intriguingly, one defense to the "judicial accountability" critics is that judges on an appellate panel who disagree with the court's decision will provide a concurring or dissenting opinion, and the majority and minority opinions would then qualify for publication.(5) However, significant numbers of cases with concurring or dissenting opinions go unpublished. Thus, even in cases where appellate judges disagree about the law or its application to the facts the court nevertheless decides not to make this dispute public. Moreover, it is possible that an unprincipled but unanimous court might decide not to publish its opinion.

Critics of limited publication plans have also argued that limited publication hampers judicial review, as unpublished opinions could be used to hide embarrassing information about litigants or their counsel,(6) or to send messages to government agencies on future cases without disapproving of past cases.(7) Moreover, the prospects for the further review of unreported appellate panel decisions are diminishing. For example, in 1995 the U.S. Courts of Appeals issued opinions in 30,240 cases, 22,054 of which were unpublished. By contrast, in the 1995 Term the Supreme Court issued opinions in sixty-four cases arising from the circuits, only two of which were unpublished dispositions.(8) The growing caseload of the circuits has resulted in "making the 13 intermediate appellate courts more and more the nation's courts of last resort."(9)

Critics also argue that judges are poor predictors of whether their decisions will be useful as precedent, and even advocates of limited publication usually concede this point. My research has concluded that the variation in publication rates among the circuits cannot be explained by the local rules, suggesting that the circuits may not be following their own rules. This practice adds to the mounting evidence that there is a small but significant problem with courts and litigants relying on unpublished decisions despite rules to the contrary.(10)

Equality of access is also a substantial factor identified by critics of limited publication rules.(11) The use of computer technology by even solo practitioners may level the playing field, although several circuits refuse to provide unpublished opinions to computer-assisted legal research firms.(12)

The "Judicial and Litigant Economy" criticisms of limited publication suggest that there is little evidence that non-publication rules result in added productivity of the judiciary. However, analysis of data collected from each case should be useful in answering whether non-publication does save time. I have concluded that cases with unpublished opinions are resolved substantially faster than those in which the opinion is published. Moreover, decisions without publication are resolved more quickly than published decisions in every category analyzed: the action taken by the court, the subject matter of the appeal, the type of appeal, whether the government is a party, the circuit in which the appeal is heard, and whether the case is heard on oral argument or decided on briefs alone.

The Data

The number of unreported opinions has increased just as dramatically as the growth of appellate litigation. In 1981, the courts decided 18,885 appeals and just under half of decisions were unpublished(13), but in 1995, over seventy percent of the 30,240 opinions of the circuits were unpublished dispositions. Table 1 sets forth the total numbers of published and unpublished decisions from 1992 to 1995, demonstrating the overall trend toward increased use of unreported decisions. It is apparent that one mechanism the circuits have used to cope with an increasing caseload is the increased use of limited publication rules.

Table 1: Publication by Year




Percent Unpublished
1992 11811 22331 34142 65.41%
1993 8927 21526 30453 70.69%
1994 8538 20165 28703 70.25%
1995 8186 22054 30240 72.93%

Table 2 sets forth data on the number of cases published by the action taken on the merits. As one might expect, more than ninety percent of all cases dismissed as frivolous resulted in an unpublished decision. But the courts have frequently resorted to unpublished opinions in cases in which the lower court was reversed or remanded, in whole or in part. In 8,583 cases, the appellate court reversed or remanded, at least in part, yet the opinion of the appellate court was not published. These cases represent decisions where a reviewing court disagreed with the lower court, and therefore are decisions which under most circuit rules probably should have been published.(14) As Table 2 also indicates, these reversals were accomplished more rapidly than in cases with published opinions, and the dismissals of frivolous appeals occurred considerably sooner when the opinion was not published. While the mean number of days pending(15) for appeals with published affirmances was 450.33, unpublished affirmances occurred in just 325.81 days. The disparity was somewhat less for outright reversals, with published opinions issued 455.63 days after the record was complete, compared to 405.94 days for unpublished opinions. By contrast, unpublished dismissals of frivolous appeals occurred after a mean of 221.36 days, or less than half the time of published reversals.

Table 2: Publication and Mean Days Pending

by Action on the Merits

Publication Mean Days Pending
Action on the Merits


Percent Unpublished


All Cases
Affirmed or Enforced 19521 65328 76.99% 450.33 325.81 362.16
Reversed or Vacated 7563 4543 37.53% 455.63 405.94 436.98
Affirmed in Part and

Reversed in Part






Dismissed - Frivolous 460 5384 92.13% 382.46 221.36 234.04
Remanded 886 1350 60.38% 486.78 363.10 412.11
Other 514 952 64.94% 489.29 411.68 438.89
ALL CASES 34348 80247 70.03% 464.88 337.20 375.47

Table 3 indicates that there were 1,063 cases in which a separate concurring or dissenting opinion was filed. This relatively low figure suggests that publication of all cases in which a member of the panel disagrees with the majority would not be an undue burden on publishers or the legal community. Indeed, inclusion of these 1,063 cases would only increase the total number of published opinions by less than three percent. Moreover, as also demonstrated by Table 3, the time costs of publication do not appear to be significant. Although unpublished opinions in which there was a concurrence were resolved ninety days sooner than published cases, unpublished cases in which a dissent was filed were resolved only seven days sooner than published cases with a dissenting opinion. Thus, publication of all cases in which there is a dissenting or concurring opinion would come at little cost, either in time or volume.

Table 3: Publication and Mean Days Pending

by Separate Opinions

Publication Mean Days Pending






All Cases
No separate opinion 33021 88689 72.87% 402.61 306.22 334.61
Dissenting opinion 2917 641 18.02% 457.22 449.92 455.76
Concurring opinion 1364 400 22.68% 467.36 377.00 448.06
Both concurring and dissenting opinion






Both concurring and dissenting opinion

- en banc






ALL CASES 37587 87752 70.01% 409.46 307.88 340.31

Table 4 presents data on publication rates and mean days pending by type of appeal. Considerable variation exists in publication rates among the different types of appeals. Notably, nearly seventy percent of all civil private cases were resolved without a published opinion;(16) three-quarters of civil cases involving the United States were not published, and two-thirds to four-fifths of cases under the federal sentencing guidelines were unpublished dispositions.

Table 4 also demonstrates that in nearly every type of appeal unpublished cases were completed more quickly than those cases with published opinions, and unpublished civil cases were handled as much as one hundred fewer days than published cases. Unpublished criminal cases (cases arising prior to the adoption of sentencing guidelines) pending for a mean of 1,821 days can only be attributed to a statistical anomaly due to only fifteen cases falling into this category.

Table 4: Publication and Mean Days Pending by Type of Appeal
Publication Mean Days Pending
Type of Appeal


Percent Unpublished


All Cases
Admin. Review 2742 4304 61.08% 400.29 314.38 350.71
Admin. Enforcemt.






Civil, U.S. 4959 14922 75.06% 402.57 294.40 325.60
Civil, Private 16511 38245 69.85% 408.47 302.76 338.17
Criminal 35 15 30.00% 877.91 1821.42 1118.81
Original Proceeding 163 2307 93.40% 383.86 317.00 363.80
Bankruptcy, from BAP 94 187 66.55% 568.38 502.15 524.30
Bankruptcy, from Dist.






Pre-guidelines 1467 2447 62.52%



Guidelines - general






Guidelines - sentence only






Guidelines - conviction only






Guidelines - sentence & conviction






ALL CASES 37587 87752 70.01% 409.46 307.88 340.31

Table 5 shows that wide variation in publication rates continues to exist among the circuits, ranging from 41.41% of cases unpublished in the First Circuit to 84.19% of cases unpublished in the Fourth Circuit. Table 5 also reveals that there are similar variations in the mean days pending for appeals, with a high of 408.19 days in the Seventh Circuit to a low of 215.33 days in the Second Circuit. In every circuit unpublished cases are resolved more quickly than cases with published opinions, and in some circuits, such as the Ninth and Eleventh, cases with published opinions can take substantially longer to resolve. Further research may demonstrate whether this variation can be explained by variation in the circuits as to judicial workload, presence or absence of senior or visiting judges, or the number of vacant judgeships.

Table 5: Publication and Mean Days Pending by Circuit

Publication Mean Days Pending


Percent Unpublished Published Unpublished All Cases
D.C. Circuit 1985 2391 54.64% 354.52 268.73 324.73
1st Circuit 2063 1458 41.41% 316.71 220.52 277.90
2nd Circuit 3125 5263 62.74% 264.74 184.58 215.33
3rd Circuit 1885 7089 78.99% 325.43 203.88 231.23
4th Circuit 1946 10363 84.19% 354.49 291.75 309.38
5th Circuit 4728 12457 72.49% 360.10 234.17 271.69
6th Circuit 2268 8211 78.36% 425.28 273.30 307.09
7th Circuit 4770 3806 44.38% 436.53 371.58 408.19
8th Circuit 4620 5354 53.68% 364.72 261.59 331.41
9th Circuit 4750 16265 77.40% 576.09 435.14 467.37
10th Circuit 2765 4917 64.01% 385.72 249.67 307.79
11th Circuit 2682 10178 79.14% 521.12 338.35 376.70
ALL CASES 37587 87752 70.01% 409.46 307.88 340.31

The data clearly demonstrate that cases selected for oral argument are more frequently those in which a published opinion will issue. Whether a case is heard on oral argument does not determine whether an opinion will be published. Indeed, it is somewhat surprising that decisions in 44.37% of cases judged significant enough to merit oral argument are not deemed worthy of publication. It is clear that oral argument increases the amount of time a case remains pending. Given the need for notice to the parties and time for the court and litigants to prepare for oral argument, it is unsurprising that cases which are argued take an average of over 100 days longer to resolve than those which are submitted solely on briefs. In contrast, lack of publication saves only 55 days for cases which are heard on oral argument. Courts would apparently save considerably more time by selecting fewer cases for oral argument rather than reducing publication rates.

Table 6: Publication & Mean Days Pending by Oral Argument

Publication Mean Days Pending


Percent Unpublished Published Unpublished All Cases
Argued 26801 21375 44.37% 415.46 360.27 390.98
Not Argued 2661 41465 93.97% 349.02 280.88 284.99
ALL CASES 29462 62840 68.08% 409.46 307.88 340.31

As noted above, when the circuits originally adopted limited publication plans, there was considerable variation in the criteria used to determine whether a decision should be published, how many votes are required to publish an opinion, and whether unpublished opinions may be cited. These variations persist today and have resulted in significant disparities in publication rates. Four circuits have a presumption in favor of publication of opinions, and six have a presumption against publication. Criteria for publication range from specific to vague, or none at all. In some circuits one member of a panel can decide that the opinion should be published, and in others a majority is required. Most courts disfavor the citation of unreported opinions, and some prohibit citation.(17)

Some evidence from other researchers suggests that these rules are not always followed, usually through review of the content of the unreported opinions to determine whether they make new law. Careful comparison of publication rates by circuit with each court's local rules regarding presumptions about publication provides additional evidence that the criteria in the courts' local rules may not always be followed. Two circuits with rules favoring publication nevertheless maintain high rates of unpublished decisions. Over 78% of decisions go unpublished in the Sixth Circuit, while 72.49% of decisions were not published in the Fifth Circuit, two courts with rules favoring publication of opinions. The Seventh Circuit has a rule disfavoring publication, yet at 44.38% had the second lowest rate of non-publication. Thus, although the local rules state each circuit's policy regarding publication, the data suggest that these policies are not uniformly implemented.

Recommendations and Conclusion

The data I have studied demonstrate that significant time savings occur through the use of limited publication plans by the appellate courts. With each variable examined, unpublished cases were concluded more quickly than published cases, and in some cases several months sooner. The courts certainly use limited publication to resolve routine cases more quickly. The data show that many decisions on prisoner petitions and administrative appeals are not published, but at the same time large numbers of cases in non-routine categories also go unpublished.

The data also indicate that a rule requiring the publication of cases with dissents or concurrences would not be a substantial burden, increasing the total number of published cases over four years by only 1,063. A significantly higher number of cases would be added to publication if all reversals were published, adding 8,523 cases to the 32,348 decisions published during the fifty-two months of opinions studied in this research. The benefits of publication of cases with separate opinions and of reversals outweigh the burden of publication. Such a rule would make public judicial expositions about the law or its application to the facts in cases in which reasonable minds may differ. This modification of the rules could quell criticism of limited publication rules on the basis of judicial accountability at minimal cost to judicial efficiency. Finally, revision of publication rules would help the public and legal community understand what appellate judges are thinking about the evolution of the law.


1. Committee on the Use of Appellate Energies, Advisory Council for Appellate Justice, Standards for Publication of Judicial Opinions (1973). For a more detailed explanation of the Advisory Committee report, see Robert J. Martineau, Restrictions on Publication and Citation of Judicial Opinions: A Reassessment, 28 U. Mich. J.L. Ref. 119 (1995) (hereinafter "Martineau").

2. William L. Reynolds & William M. Richman, The Non-precedential Precedent - Limited Publication and No-Citation Rules in the United States Courts of Appeals, 78 Colum. L. Rev. 1167 (1978) (hereinafter "Reynolds & Richman I"). See also, William L. Reynolds & William M. Richman, An Evaluation of Limited Publication in the United States Courts of Appeals: The Price of Reform, 48 U. Chi. L. Rev. 573 (1981) (hereinafter "Reynolds & Richman II"); William L. Reynolds & William M. Richman, Limited Publication in the Fourth and Sixth Circuits, 1979 Duke L.J. 807, 809; Milton J. Silverman, The Unwritten Law: The Unpublished Opinion in California, 51 Cal. St. B.J. 33 (1976); Pamela Foa, Comment, A Snake in the Path of the Law: The Seventh Circuit's Non-Publication Rule, 39 U. Pitt. L. Rev. 309 (1977); Donald R. Songer, Danna Smith and Reginald S. Sheehan, Nonpublication in the Eleventh Circuit: An Empirical Analysis, 16 Fla. St. U.L. Rev. 964 (1989).

3. Martineau, supra n. 2, at 128-145.

4. Martineau, supra n. 2, at 130. By their very nature, unreported decisions are not usually easily accessible to the public, so access to data about unreported decisions has been difficult for researchers not affiliated with the courts.

5. Id. at 129.

6. Judge Nichols of the Federal Circuit reports that in his experience as a visiting judge the Fifth and Eleventh Circuits will issue an opinion which reads in its entirety "AFFIRMED. See Loc.R. 47.6" in cases where "the case presents no genuine appealable issue and the parties who initiated the appeal should have known this and probably did", which he terms "a rebuke for misuse of the appellate process, but one administered with true Southern courtesy." Philip Nichols, Jr., Selective Publication of Opinions: One Judge's View, 55 Am.U. L. Rev. 909 (1986).

7. Id. at 133. Table 6 sets forth data regarding unreported cases in which the government is a party.

8. The two cases are Behrens v. Pelletier, ___ U.S. ___, 116 S. Ct. 834, 133 L. Ed. 2d 773 (1996) (reversing an unpublished Eleventh Circuit opinion) and Things Remembered, Inc. v. Petrarca, ___ U.S. ___, 116 S. Ct. 494, 133 L. Ed. 2d 461 (1995) (affirming an unpublished Sixth Circuit opinion). In addition, the Court noted in Commissioner of Internal Revenue v. Lundy, ___ U.S. ___, ___, 116 S. Ct. 647, 650 (1996) that it granted certiorari to resolve a conflict between the Fourth Circuit's reported decision and a reported decision of the Tenth Circuit and unreported decisions from the Second, Sixth and Ninth Circuits. During the 1996 Term, the Court has reviewed two unpublished opinions of the circuits, Lynce v. Mathis, ___ U.S. ___, 117 S. Ct. 891 (1997) (reversing an unpublished Eleventh Circuit opinion) and Old Chief v. United States, ___ U.S. ___, 117 S. Ct. 644 (1997) (reversing an unpublished Ninth Circuit opinion).

9. Federal Judicial Center, Report of the Federal Courts Study Committee 110 (1990). The Report notes that while the Supreme Court granted certiorari in 157 of the 1,992 appellate cases decided in 1945 (7.9% of all decisions of the circuits), the 142 grants of certiorari in the 19,178 cases decided by the circuits in 1989 amounted to review of less than one percent of the appellate court's decisions.

10. See, e.g., Foa, supra n. 3; Lauren K. Robel, The Myth of the Disposable Opinion: Unpublished Opinions and Government Litigants in the United States Courts of Appeals, 87 Mich. L. Rev. 940 (1989); Songer, Smith & Sheehan, supra n. 3; Peter J. Honigsberg & James A. Dikel, Unfairness in Access to and Citation of Unpublished Federal Court Decisions, 18 Golden Gate U. L. Rev. 277 (1988).

11. See, e.g., Foa, supra n. 3; Robel, supra n. 11; Reynolds & Richman I, supra n. 3; Peter J. Honigsberg & James A. Dikel, Unfairness in Access to and Citation of Unpublished Federal Court Decisions, 18 Golden Gate U. L. Rev. 277 (1988).

12. The Second, Third, Fifth and Eleventh Circuits do not make unreported opinions available to electronic publishers. Martineau, supra n. 2, at 144.

13. Donna Steinstra, Unpublished Dispositions: Problems of Access and Use in the Courts of Appeals at 3 (1985) (finding that 48.8% of decisions in 1981 were not published). Data on the numbers of unpublished opinions have been reported in the Annual Reports of the Administrative Office only since 1989. One study indicates that non-publication rates in the 1960s were low as 16.3%. J. Woodford Howard, Jr., Courts of Appeals in the Federal Judicial System: A Study of the Second, Fifth and District of Columbia Circuits (1981).

14. But Judge Nichols notes that in some cases his concurrence or dissent might not be based upon issues of law. Nichols, supra n. 7 at 925. Moreover, Judge Nichols notes that publication of a decision in which he dissents would make the majority decision a binding precedent, a result many dissenters wish to avoid. Id. On the other hand, Robel indicates that mere information about the opinion of an individual judge may be valuable to litigants. Robel, supra n. 11 at 947.

15. The mean days pending was calculated from the date the record was complete until the date of final judgment, rather than from the date the appeal was filed, so that delays in assembling the lower court's record would not be included.

16. "Civil, private cases" includes many cases which fall into the "prisoner petition" category brought not as criminal appeals but as civil actions against state officials.

17. Courts prohibiting citation usually permit citation to establish law of the case, res judicata or collateral estoppel. See, e.g., 9th Cir. R. 36-3.