PUBLISHED

UNITED STATES COURT OF APPEALS

FOR THE FOURTH CIRCUIT

GARY LEE ROLLER,
Plaintiff-Appellant,

v.

WILLIAM E. GUNN, Executive
Director of the South Carolina
Department of Probation, Parole and
                                                                      No. 96-6992
Pardon Services; SOUTH CAROLINA
DEPARTMENT OF PROBATION, PAROLE
AND PARDON SERVICES,
Defendants-Appellees,

UNITED STATES OF AMERICA,
Intervenor.

Appeal from the United States District Court
for the District of South Carolina, at Columbia.
Henry M. Herlong, Jr., District Judge.
(CA-90-3054-3-20-J)

Argued: December 2, 1996

Decided: February 19, 1997

Before WILKINSON, Chief Judge, and RUSSELL and HALL,
Circuit Judges.

_________________________________________________________________

Affirmed by published opinion. Chief Judge Wilkinson wrote the
opinion, in which Judge Russell joined. Judge Hall wrote an opinion
concurring in part and dissenting in part.

_________________________________________________________________
COUNSEL

ARGUED: Rochelle Romosca McKim, W. Gaston Fairey, FAIREY,
PARISE & MILLS, P.A., Columbia, South Carolina, for Appellant.
Carl Norman Lundberg, Chief Legal Counsel, SOUTH CAROLINA
DEPARTMENT OF PROBATION, PAROLE AND PARDON SER-
VICES, Columbia, South Carolina, for Appellees. Richard Alan
Olderman, Appellate Staff, Civil Division, UNITED STATES
DEPARTMENT OF JUSTICE, Washington, D.C., for Intervenor.
ON BRIEF: Charles Molony Condon, Attorney General, Donald J.
Zelenka, Assistant Deputy Attorney General, Reginald I. Lloyd,
Assistant Attorney General, OFFICE OF THE ATTORNEY GEN-
ERAL, Columbia, South Carolina, for Appellees. Frank W. Hunger,
Assistant Attorney General, J. Rene Josey, United States Attorney,
Barbara L. Herwig, Appellate Staff, Civil Division, UNITED
STATES DEPARTMENT OF JUSTICE, Washington, D.C., for Inter-
venor.

_________________________________________________________________

OPINION

WILKINSON, Chief Judge:

Appellant, Gary Lee Roller, challenges the constitutionality of sec-
tion 804 of the Prison Litigation Reform Act ("PLRA"), Pub. L. No.
104-134, 110 Stat. 1321, §§ 801-810 (Apr. 26, 1996), which requires
prisoners proceeding in forma pauperis to pay a partial filing fee
before filing a lawsuit or proceeding with an appeal. We think the Act
represents a legitimate exercise of Congress' power to reduce frivo-
lous lawsuits in the federal courts. We reject Roller's contentions that
the filing fee requirement imposes an unconstitutional barrier on
access to the courts or that it violates the Constitution's equal protec-
tion guarantee.

Roller also appeals the district court's conclusion that the retro-
spective application of amendments to South Carolina's parole proce-
dures does not constitute a violation of the Ex Post Facto Clause. The
Supreme Court's recent decision in California Dept. of Corrections
v. Morales, 

(1995), permits the states to adopt

                     2
changes to their parole procedures which produce no more than a
speculative possibility of affecting a prisoner's punishment. Under
Morales, the changes to South Carolina's parole procedures do not
violate the Ex Post Facto Clause. Accordingly, we affirm the judg-
ment of the district court.

I.

In early 1983, Gary Lee Roller was convicted of voluntary man-
slaughter and grand larceny in South Carolina state court for crimes
committed on December 13, 1982. On March 25, 1983, he was sen-
tenced to thirty-five years imprisonment.

In December 1990, Roller filed a complaint under 42 U.S.C.
§ 1983 challenging the application of amendments to the parole laws
of South Carolina on ex post facto grounds. The amendments, passed
in 1986, modified South Carolina Code section 24-21-645 so that a
prisoner convicted of committing a violent crime would only have his
case reviewed every two years after an initial negative parole determi-
nation rather than every year. Additionally, the amendments required
a two-thirds majority of the parole board to authorize parole for vio-
lent offenders rather than a simple majority. By its terms, the
amended version of section 24-21-645 applied to Roller despite the
fact that he had committed his crimes and been convicted prior to its
enactment.

The district court entered judgment for the South Carolina Depart-
ment of Probation, Parole, and Pardon Services, but was reversed by
this court in Roller v. Cavanaugh, 

(4th Cir. 1993). The
case was remanded with instructions to grant declaratory relief in
favor of Roller.

In April 1995, the Supreme Court decided California Dept. Of Cor-
rections v. Morales, 

(1995). Morales held that the
retroactive application of a California statute which changed the fre-
quency of parole hearings for multiple murderers from every year to
once every three years did not violate the Ex Post Facto Clause. In
January 1996, South Carolina, arguing that Morales represented a sig-
nificant change in the law, moved under Fed. R. Civ. P. 60(b) for
modification of the court's order declaring the retrospective applica-

                    3
tion of the 1986 amendments to section 24-21-645 unconstitutional.
On June 5, 1996, the district court concluded on the basis of Morales
that "the application of § 24-21-645 to Roller does not violate the Ex
Post Facto Clause." Roller v. Gunn, 

, 730 (D.S.C.
1996). Roller then filed the instant appeal.

On appeal, Roller moved this court to declare the filing fee require-
ment of the PLRA and a similar requirement providing for the pay-
ment of costs unconstitutional. Roller has paid the full amount of the
$105 filing fee under protest pending our decision.

II.

A.

We turn initially to Roller's contentions that the filing fee provi-
sions of the PLRA constitute an unconstitutional barrier to access to
the courts and violate the constitutional guarantee of equal protection.

Congress has long regulated the access of indigent litigants to the
federal judicial system. The first federal in forma pauperis ("IFP")
statute was enacted in 1892. Act of July 20, 1892, ch. 209, 27 Stat.
252 (codified as amended at 28 U.S.C. § 1915). Concerned that citi-
zens were being denied their day in court, Congress proposed to
"open the United States courts to a class of American citizens who
have rights to be adjudicated, but are now excluded practically for
want of sufficient money or property to enter the courts under their
rules." H.R. Rep. No. 1709, 52d Cong., 1st Sess. 1 (1892). Despite
Congress' admonition that "[t]he proposed law will not admit of vex-
atious litigation," 

noble purpose has been threatened
by a flood of meritless lawsuits.

Unsurprisingly, prisoners proved responsible for much of this liti-
gation. In some instances, individual prisoners have filed an astonish-
ing number of frivolous lawsuits. See, e.g., Shieh v. Kakita, 

(1996) (prisoner prospectively barred from filing petitions for
certiorari in forma pauperis after filing 10 frivolous petitions in less
than three years); In re McDonald, 

, 184 (1989) (pris-
oner prospectively barred from using in forma pauperis procedures to

                     4
file extraordinary writs after 73 frivolous filings between 1971 and
1989); In re Green, 

, 781 (D.C. Cir. 1981) (prisoner filed
"between 600 and 700 complaints in the federal and state courts"). In
1995, prisoners brought over 25% of the civil cases filed in the fed-
eral district courts. Administrative Office of the United States Courts,
1995 Federal Court Management Statistics 167. In this circuit alone,
IFP filings accounted for almost half of the court's 1995 caseload,
Nasim v. Warden, Maryland House of Correction, 

, 954
n.2 (4th Cir. 1995) (en banc), and prisoners were responsible for 75%
of those filings. 

n.1.

Congress recognized that the explosion of IFP litigation presents
problems for our legal system. It was obviously concerned that the
limited resources of the federal judiciary not be expended on cases
whose frivolity was manifest, but whose sheer numerosity represented
a formidable and time consuming task. As the Supreme Court has
noted, "[t]he goal of fairly dispensing justice . . . is compromised
when the Court is forced to devote its limited resources to the pro-
cessing of repetitious and frivolous requests." In re Sindram, 

, 179-80 (1991).

Finding that the proliferation of prisoner litigation was due signifi-
cantly to the lack of economic disincentives to filing meritless cases,
Congress passed the Prison Litigation Reform Act, Pub. L. No. 104-
134, 110 Stat. 1321, §§ 801-810 (Apr. 26, 1996). The goal of the
PLRA amendments to the in forma pauperis statute, 28 U.S.C.
§ 1915, is straightforward:

          Section 2 will require prisoners to pay a very small share of
          the large burden they place on the Federal judicial system
          by paying a small filing fee upon commencement of law-
          suits. In doing so, the provision will deter frivolous inmate
          lawsuits. The modest monetary outlay will force prisoners
          to think twice about the case and not just file reflexively.
          Prisoners will have to make the same decision that law-
          abiding Americans must make: Is the lawsuit worth the
          price?

141 Cong. Rec. at S7526 (May 25, 1995) (statement of Senator Kyl)
(citation omitted).

                    5
The PLRA thus makes several changes to 28 U.S.C.§ 1915. The
Act requires prisoners seeking IFP status to execute an affidavit
attesting to their impoverishment and to file with the court a certified
copy of their prison trust account. 28 U.S.C. § 1915(1) & (2). More-
over, under the PLRA amendments,

          if a prisoner brings a civil action or files an appeal in forma
          pauperis, the prisoner shall be required to pay the full
          amount of a filing fee. The court shall assess, and when
          funds exist, collect, as a partial payment of any court fees
          required by law, an initial partial filing fee of 20 percent of
          the greater of --

           (A) the average monthly deposits to the prisoner's
          account; or

           (B) the average monthly balance in the prisoner's
          account for the 6-month period immediately preceding the
          filing of the complaint or notice of appeal.

          (2) After payment of the initial partial filing fee, the pris-
          oner shall be required to make monthly payments of 20 per-
          cent of the preceding month's income credited to the
          prisoner's account. The agency having custody of the pris-
          oner shall forward payments from the prisoner's account to
          the clerk of the court each time the amount in the account
          exceeds $10 until the filing fees are paid.

28 U.S.C. § 1915(3)(b)(1) & (b)(2). The amendments also state that
"[i]n no event shall a prisoner be prohibited from bringing a civil
action or appealing a civil or criminal judgment for the reason that the
prisoner has no assets and no means by which to pay the initial partial
filing fee." 28 U.S.C. § 1915(4).

In addition, the PLRA amendments provide that if a district court
in its discretion awards a judgment against a prisoner and includes
costs, the prisoner "shall be required to pay the full amount of the
costs ordered" and "shall be required to make payments for costs
under this subsection in the same manner as is provided for filing fees
. . . ." 28 U.S.C. § 1915(f)(2)(A) & (B).

                     6
B.

Roller first contends that the PLRA amendments to 28 U.S.C.
§ 1915 constitute a violation of his constitutional right of access to the
courts. See Bounds v. Smith, 

(1977). He argues that by
imposing liability for fees and costs on inmates wishing to pursue
appeals, the PLRA amendments have the practical effect of foreclos-
ing access for those prisoners who are unable or unwilling to assume
the financial burdens imposed by the statute.

Roller's mistakes are fundamental. To begin with, the right of
access to federal courts is not a free-floating right, but rather is sub-
ject to Congress' Article III power to set limits on federal jurisdiction.
Indeed, Congress is neither "constitutionally required to create Article
III courts to hear and decide cases within the judicial power of the
United States," nor to vest those courts that are created "with all the
jurisdiction it was authorized to bestow under Article III." Palmore
v. United States, 

, 400-01 (1973). Congress is no more
compelled to guarantee free access to federal courts than it is to pro-
vide unlimited access to them. The Supreme Court has never recog-
nized an "unlimited rule that an indigent at all times and in all cases
has the right to relief without the payment of fees." United States v.
Kras, 

, 450 (1973). If we were to adopt Roller's argu-
ment, all filing fees would be unconstitutional, which, of course, they
are not. "The correct principle is that reasonable costs may be
imposed on persons who want to sue." Lumbert v. Illinois Dept. of
Corrections, 

, 259 (7th Cir. 1987). 1
_________________________________________________________________
1 The Supreme Court has struck down filing fee requirements in certain
state court cases involving domestic disputes. However, these cases
address situations where a filing fee presents an insurmountable barrier
to the protection of certain fundamental rights. For example, in M.L.B.
v. S.L.J., 

(U.S. Dec. 16, 1996), the Supreme Court
held that Mississippi could not condition an appeal from the termination
of parental rights on the payment of a $2,532.36 record preparation fee.
Similarly, in Boddie v. Connecticut, 

(1971), the Court held
that Connecticut could not deny a married couple access to divorce pro-
ceedings due to an inability to pay court fees. However, the Court has
made clear that in "the mine run of cases" which do not involve "state
controls or intrusions on family relationships" filing fees may be
required. See 

.

                     7
Indeed, this Court has already upheld the practice of requiring pris-
oners to pay filing fees. In Evans v. Croom, 

(4th Cir.
1981), we approved a district court rule requiring partial filing fees
in prisoner IFP lawsuits. The rule that we approved bears many simi-
larities to the scheme adopted by the PLRA. The district court in
Evans had instructed the clerk to obtain a certified copy of each plain-
tiff prisoner's trust fund account for the six month period preceding
the lawsuit. The court further instructed the clerk to require payment
of a partial filing fee not to exceed 15% of the amount of money
received in the trust fund account during the previous six month
period. We upheld the district court's order, observing that prisoners
had no financial disincentives to litigate, hence"nothing to lose and
everything to gain" from suing. 

We also noted that the dis-
trict court's order simply forced the prisoner to weigh whether the
"merit of the claim" was worth "the cost of pursuing it." 

(citation omitted).

We recently reaffirmed Evans in Nasim v. Warden, Maryland
House of Correction, 

(4th Cir. 1995) (en banc), observ-
ing that some financial accountability could be built into the system
that would retain access to the courts "without overwhelming the effi-
cient administration of justice with meritless cases." 

& n.3.
Nine other circuits have upheld the imposition of partial filing fees on
IFP plaintiffs. See In re Stump, 

, 1298 (1st Cir. 1971);
In re Epps, 

, 967 (2d Cir. 1989); Bullock v. Suolmela,

, 103 (3d Cir. 1983); Smith v. Martinez, 

,
574 (5th Cir. 1983); Clark v. Ocean Brand Tuna , 

, 50 (6th
Cir. 1992); Bryan v. Johnson, 

, 458 (7th Cir. 1987); In
re Williamson, 

, 1339-41 (8th Cir. 1986); Olivares v.
Marshall, 

, 111 (9th Cir. 1995); Collier v. Tatum, 

, 655 (11th Cir. 1983).

As meritless as his filing fee contention is Roller's claim that he
is denied access to the courts by the PLRA provision requiring pay-
_________________________________________________________________
Roller's case is not implicated by M.L.B. or Boddie. It clearly does not
involve "state controls or intrusions on family relationships." It also does
not present the sort of insurmountable barrier to filing suit considered in
those cases.

                    8
ment of costs assessed against a prisoner, 28 U.S.C.§ 1915(f)(2). In
Flint v. Haynes, 

(4th Cir. 1981), this circuit held that
"a district court is empowered to award costs even when it has previ-
ously granted a litigant the benefits" of IFP status. 

We
went on to reject the argument that requiring payment of costs
infringed an IFP litigant's right of access to the courts:

          [T]he appellants argue that in order to insure that indigent
          civil rights litigants have access to the courts to redress their
          grievances, costs should rarely be assessed in these circum-
          stances. However, when costs are assessed only in extreme
          or exceptional cases, those persons granted leave to proceed
          in forma pauperis have virtually nothing to lose and every-
          thing to gain, and the purpose of § 1915 -- equal access for
          the poor and rich -- is distorted. Non-indigents who con-
          template litigation are routinely forced to decide whether
          their claim is worth it. We see no reason to treat indigents
          differently in this respect.

(citations omitted); see also Weaver v. Toombs, 

, 1008 (6th Cir. 1991); Harris v. Forsyth , 

, 1277-
78 (11th Cir. 1984).

In effect, the PLRA simply follows the course set by the federal
courts in cases like Evans v. Croom and Flint v. Haynes. The statute's
approach is hardly draconian. Section 1915 requires a modest initial
filing fee before a case may proceed, 28 U.S.C.§ 1915(b)(1), and
thereafter a prisoner is required to pay only "20 percent of the preced-
ing month's income credited to the prisoner's account" until the total
fee is paid, 28 U.S.C. § 1915(b)(2). To further ensure that prisoners
need not "totally deprive themselves of those small amenities of life
which they are permitted to acquire in a prison or mental hospital
beyond the food, clothing, and lodging already furnished by the
state," 

(citation omitted), section 1915 allows
payment to be taken from the prisoner's account only where "the
amount in the account exceeds $10 . . . ." 28 U.S.C. § 1915(b)(2).
Furthermore, the PLRA provides that "[i]n no event shall a prisoner
be prohibited from bringing a civil action or appealing a civil or crim-
inal judgment for the reason that the prisoner has no assets and no
means by which to pay the initial partial filing fee." 28 U.S.C.

                     9
§ 1915(b)(4). These mild steps do not begin to impose an unconstitu-
tional burden on a prisoner's access to the courts.

Roller maintains that even though the PLRA amendments include
safeguards to ensure the truly indigent will have access to the courts,
the practical effect of the amendments will be that prisoners with
other expenses will be unable to afford lawsuits. This argument, how-
ever, misses the entire point of the statute. Requiring prisoners to
make economic decisions about filing lawsuits does not deny access
to the courts; it merely places the indigent prisoner in a position simi-
lar to that faced by those whose basic costs of living are not paid by
the state. Those living outside of prisons cannot file a lawsuit every
time they suffer a real or imagined slight. Instead, they must weigh
the importance of redress before resorting to the legal system. If a
prisoner determines that his funds are better spent on other items
rather than filing a civil rights suit, "he has demonstrated an implied
evaluation of that suit" that the courts should be entitled to honor.

.

C.

Roller contends finally that the PLRA amendments violate the
Constitution's guarantee of equal protection. Prisoners are not a sus-
pect class. See Pryor v. Brennan, 

, 923 (7th Cir. 1990);
Moss v. Clark, 

, 690 (4th Cir. 1989). Nor is indigency
a suspect classification. See Harris v. McRae , 

, 323
(1980); Maher v. Roe, 

, 471 (1977). Moreover, as we
have explained above, the PLRA amendments do not burden any fun-
damental rights. We therefore review the PLRA amendments under
a rational basis standard. "Unless a classification trammels fundamen-
tal personal rights or is drawn upon inherent suspect distinctions such
as race, religion, or alienage, our decisions presume the constitution-
ality of the statutory discrimination and require only that the classifi-
cation challenged be rationally related to a legitimate state interest."
City of New Orleans v. Dukes, 

, 303 (1976).

The PLRA amendments easily satisfy the rational basis standard.
Congress has acted in an area in which it has important responsibili-
ties. As the Supreme Court recognized in Hanna v. Plumer, 

(1965), "the constitutional provision for a federal court system

                     10
(augmented by the Necessary and Proper Clause) carries with it con-
gressional power to make rules governing the practice and pleadings
in those courts . . . ." 

Moreover, the goal of the Act --
curbing frivolous IFP litigation -- is clearly proper. We have earlier
discussed the legitimacy of this interest, and we shall not repeat that
discussion here.

Roller, however, maintains that the means Congress used to further
its clearly legitimate goal are irrational, contending that Congress
impermissibly singled out the prison population for the filing fee
requirement.2 We disagree. There are a host of rational grounds for
imposing the filing fee requirement on prisoners."The Constitution
does not require things which are different in fact . . . to be treated
in law as though they were the same." Tigner v. Texas, 

,
147 (1940). Prisoners are not similarly situated to non-prisoners. They
have their basic material needs provided at state expense. They are
further provided with free paper, postage, and legal assistance. They
often have free time on their hands that other litigants do not possess.
See 

. As a result, the federal courts have
observed that prisoner litigation has assumed something of the nature
of a "recreational activity." See, e.g., Gabel v. Lynaugh, 

,
125 n.1 (5th Cir. 1988). Whether recreational or not, there has been
_________________________________________________________________
2 Roller, relying primarily on Rinaldi v. Yeager, 

(1966),
contends that the PLRA amendments do not satisfy the requirements of
rationality. Roller's reliance on Rinaldi, however, is misplaced. In
Rinaldi, the Supreme Court found a New Jersey statute violated equal
protection because it required a single class of unsuccessful criminal
appellants -- imprisoned indigents -- to reimburse the state for the costs
of trial transcripts. Indigent appellants who had been convicted of a
crime and had been unsuccessful on appeal, but who had received pun-
ishments other than imprisonment, were not subject to this requirement.
The Supreme Court found there was simply no basis for this distinction.
As the Supreme Court subsequently characterized its decision, "the New
Jersey distinction . . . was invidious and without rationality for it was not
related to the fiscal objectives of the statute and rested on no administra-
tive convenience." Schlib v. Kueble, 

, 369 (1971).

Rinaldi simply is not applicable here. Congress has not singled out a
single class of appellants to pay a "penalty" if unsuccessful. Instead,
Congress has uniformly imposed a requirement on those prisoners wish-
ing to take advantage of the privilege of IFP status.

                    11
a far greater opportunity for abuse of the federal judicial system in the
prison setting. See 141 Cong. Rec. S7256 (May 25, 1995) (statement
of Sen. Kyl) (noting that over one-fourth of civil cases filed in federal
district courts were filed by prisoners, and that the vast majority of
these cases ended in no relief for the prisoner). Congress could ratio-
nally have concluded that this abuse of the federal judicial system by
inmates was likely to continue absent significant changes in the IFP
statute.

There are other reasons why Congress may have viewed the prison
setting as uniquely appropriate for a filing fee requirement. Many fed-
eral courts had adopted varied plans for prisoner filings similar to the
PLRA amendments. Congress may have wanted to establish uniform
national standards for the handling of such cases in the federal courts.
Congress may also have wanted to establish a means of ensuring that
those who claim IFP status truly are indigent. The PLRA amendments
establish a uniform method for making such determinations among
prisoners. Furthermore, because prisoners are under the control of the
state, it is administratively easier for the courts to check the finances
of inmates than other IFP plaintiffs. Congress may have limited the
filing fee requirements to prisoners because of this administrative
convenience. After all, a legislature "may take one step at a time,
addressing itself to the phase of the problem which seems most acute
to the legislative mind." Bowen v. Owens, 

, 347 (1986)
(quoting Williamson v. Lee Optical Co., 

, 489 (1955)).

In sum, the equal protection question is not a close one. The legis-
lative solution is entirely rational, does not violate any fundamental
rights, and does not single out a suspect class for disparate treatment.
We therefore hold the PLRA amendments to 28 U.S.C.§ 1915 to be
constitutional.

III.

Roller contends that the district court erred when it held the appli-
cation of section 24-21-645 constitutional. He maintains that the
Supreme Court's decision in California Dept. of Corrections v.
Morales, 

(1995), did not call into question the valid-

                     12
ity of this court's ruling in Roller v. Cavanaugh, 

(4th
Cir. 1993).3

We disagree. The Cavanaugh decision reasoned that any retroac-
tive reduction in the frequency of parole consideration would violate
the Ex Post Facto Clause because such a change would necessarily
eliminate an opportunity for early release. This is precisely the argu-
ment the Supreme Court rejected in 

;
see also Jones v. Georgia State Board of Pardons and Paroles, 

, 1149 n.8 (11th Cir. 1995) (noting that Morales called into
question the validity of Akins v. Snow, 

(11th Cir.
1991), a case which rested on the same reasoning as Cavanaugh).
Indeed, as ours and other circuits have already recognized, Morales
holds that the Ex Post Facto Clause does not foreclose every change
in state parole procedures that poses a speculative risk of increasing
a prisoner's punishment. See Artway v. New Jersey, 

,
1260-61 (3d Cir. 1996); Hamm v. Latessa, 

, 959 (1st Cir.
1995); United States v. Reese, 

, 591 (6th Cir. 1995);
Allison v. Kyle, 

, 74-75 (5th Cir. 1995); Hill v. Jackson, 

, 168 (4th Cir. 1995); 

.

In Morales, the Supreme Court considered the retroactive applica-
tion of a California statute which changed the frequency of parole
consideration for multiple murderers from every year to every three
_________________________________________________________________
3 Prior to this circuit's decision in Roller v. Cavanaugh, the South Car-
olina Supreme Court had held that section 24-21-645 did not violate the
Ex Post Facto Clauses of either the United States or South Carolina Con-
stitutions. Gunter v. South Carolina, 

, 444 (S.C. 1989).
However, in Griffin v. South Carolina, 

, 863-64 (S.C.
1993), the South Carolina Supreme Court, relying on Roller v.
Cavanaugh, reversed Gunter. Given that the Griffin decision was wholly
predicated on federal precedent, there is no independent state law ground
that would moot our consideration of this issue. As the Supreme Court
stated in Michigan v. Long, 

, 1040-41 (1983), "when, as
in this case, a state court decision fairly appears to rest primarily on fed-
eral law, or to be interwoven with the federal law, and when the ade-
quacy and independence of any possible state law ground is not clear
from the face of the opinion, we will accept as the most reasonable
explanation that the state court decided the case the way it did because
it believed that federal law required it to do so."

                    13
years. Finding that the California law involved"only the most specu-
lative and attenuated possibility of producing the prohibited effect of
increasing the measure of punishment for covered 

, the Court held that California's change in parole procedures
did not violate the Ex Post Facto Clause. While the Court declined
to express a view "as to the constitutionality of any of a number of
other statutes that might alter the timing of parole hearings," 

n.5, it did clarify the principles we must use to analyze the con-
stitutionality of changes in parole procedures.

Morales instructs that a law which changes the frequency of parole
hearings will implicate the Ex Post Facto Clause only when "it pro-
duces a sufficient risk of increasing the measure of punishment
attached to the covered crimes." 

The sufficient risk
inquiry "cannot be embraced within a formula or stated in a general
proposition," but rather is a matter of "degree." 

the legislative adjustment creates only a "speculative and attenuated"
risk of increasing punishment it will withstand constitutional attack.
Id.; Dobbert v. Florida, 

, 294 (1977).

The South Carolina law we examine today bears a strong resem-
blance to the California statute sustained in Morales. Neither the Cali-
fornia law nor the South Carolina statute increase in any way the
actual sentence of imprisonment. Like the California measure, the
South Carolina law applies only to prisoners convicted of violent
crimes -- prisoners which the South Carolina legislature determined
were unlikely to receive release on parole. See 

. As with California's law, section 24-21-645"has no effect on
the date of any prisoner's initial parole suitability hearing; it affects
the timing only of subsequent hearings." 

(emphasis in
original). Like the California provision, the South Carolina law has
not changed the substantive standards for qualifying for parole; parole
determinations remain within the total discretion of the parole board.
See 

However, unlike the California statute, which
changed the frequency of parole hearings from once a year to once
every three years, section 24-21-645 has only changed the frequency
of parole hearings to every two years.

Despite these similarities and the fact that the South Carolina stat-
ute calls for a greater frequency of parole hearings than the statute

                     14
upheld in Morales, Roller contends that Morales should be limited to
its facts. He argues that since section 24-21-645 does not include
every provision included in the California statute, the retroactive
application of the South Carolina law constitutes a violation of the Ex
Post Facto Clause. For example, the California statute requires the
parole board to make a factual finding that "it is not reasonable to
expect that parole would be granted at a hearing during the following
years," before it may defer future parole hearings. 

(citation omitted). Roller maintains that since no individual-
ized findings are required under the South Carolina law, the possibil-
ity that he might experience a change of circumstances that would
increase his chances of parole during the year his hearing was
deferred are significantly greater than under the statute sustained in
Morales. He thus contends that section 24-21-645 has increased sig-
nificantly his punishment by foreclosing for a year an opportunity for
release.

Roller's claim, however, boils down to mere speculation about his
release. Such conjecture is insufficient under Morales to establish a
violation of the Ex Post Facto Clause. In South Carolina, the determi-
nation of parole is subject to the broad discretion of the parole board.
S.C. Code Ann. § 24-21-645. Forecasts on how the board might
decide to exercise its discretion in any given case are merely in the
nature of conjecture. Roller simply fails "to provide support for his
speculation that . . . prisoners subject to [24-21-645] might experience
an unanticipated change that is sufficiently monumental to alter their
suitability for release on parole." Morales , 115 S. Ct. at 1604. Fur-
thermore, as the district court noted, there is nothing on the face of
section 24-21-645 that limits the parole board's authority to schedule
expedited hearings if presented with suitable circumstances. Roller v.
Gunn, 

, 730 (D.S.C. 1996). In Morales, this same
consideration led the Supreme Court to conclude that even if a prison-
er's circumstances drastically changed during the period that his
parole hearing had been delayed, "there is no reason to conclude that
the amendment will have any effect on any prisoner's actual term of

.

Roller also contends that section 24-21-645 effects an impermissi-
ble change in the standards for parole because it requires a two-thirds
vote of the parole board rather than the simple majority required by

                    15
the predecessor statute. This argument is foreclosed by Dobbert v.
Florida, 

(1977). In Dobbert, the trial judge overruled
the jury's recommendation of life imprisonment and imposed a sen-
tence of death. The criminal defendant claimed that the statute which
allowed a trial judge to determine finally, after a jury recommenda-
tion, whether the death sentence was appropriate violated the Ex Post
Facto Clause because under the old statute the final death penalty
determination would have been made by a majority of the jury. The
Supreme Court rejected this claim because it was based on conjecture,
stating, "[I]t certainly cannot be said with assurance that, had his trial
been conducted under the old statute, the jury would have returned a
verdict of life." 

Like the claim of the petitioner in Dobbert, Roller's claim is specu-
lative. There is no way of knowing whether a particular board mem-
ber's vote would be the same under the new two-thirds majority rule
as it would have been under the old rule. As the Supreme Court noted
in Dobbert, "[The jurors] may have chosen leniency when they knew
that that decision rested ultimately on the shoulders of the trial judge,
but might not have followed the same course if their vote were final."

n.7. Similarly, parole board members might be more likely
to vote for granting parole under the two-thirds rule, knowing that any
favorable decision must be concurred in by a greater number of their
colleagues.

Morales likewise compels us to uphold the retrospective applica-
tion of the two-thirds vote requirement. The Court specifically cau-
tioned the judiciary against the "micromanagement of an endless
array of legislative adjustments to parole and sentencing 

. If we were to invalidate the two-thirds majority
requirement, we would be ignoring the Court's admonition and open-
ing the door to a host of challenges to state parole practices. For
example, a state that allowed parole on a simple majority vote would
violate the Ex Post Facto Clause if the parole board's membership
changed from three to four members because a prisoner would have
to convince three-fourths of the board rather than two-thirds in order
to gain parole. If Morales stands for anything, it stands for the propo-
sition that this sort of judicial fine-tuning is not appropriate. See
Cavallaro v. Groose, 

, 136 (Mo. 1995) (holding that
a change from a three member board to a five member board did not

                     16
violate the Ex Post Facto Clause). The two-thirds majority require-
ment is simply a procedural change in South Carolina's parole sys-
tem, and "[e]ven though it may work to the disadvantage of a
defendant, a procedural change is not ex post facto." 

.

In the end, if we adopt Roller's arguments, we would be required
to strike down every retrospective application of parole procedures
which do not conform exactly to the California statute considered in
Morales. We have already rejected this proposition. In Hill v.
Jackson, 

(4th Cir. 1995), this court upheld a change in
Virginia's parole law which applied to a group of prisoners broader
than the class of prisoners affected by the California law upheld in
Morales. Instead of being limited to multiple murderers, the Virginia
parole policy deferred parole hearings for up to three years for those
prisoners serving life sentences or serving sentences of 70 years or
more for a violent offense. Yet we upheld the Virginia procedure, not-
ing among other things that, "[a]s in Morales, the Parole Board's pol-
icy has no effect on the substantive standards for scheduling an
inmate's initial parole eligibility date, nor does it change the criteria
for determining either an inmate's suitability for parole or his or her
release date." 

These same features characterize the South
Carolina law as well.

Our federal system does not require that every state model its
parole procedures after those of California or indeed of any other
state. "It is difficult to imagine an activity in which a State has a
stronger interest, or one that is more intricately bound up with state
laws, regulations, and procedures, than the administration of its pris-
ons." Preiser v. Rodriguez, 

, 491-92 (1973). Our federal-
ism allows states to try different solutions to the problems of
confinement and release, and to adopt diverse approaches to parole
procedures for those within state custody. The rejoinder that states
can experiment with parole procedures, but only prospectively, is no
rejoinder at all -- that position would require that states process
inmates by at least two different sets of rules in every institution.

The South Carolina statute represents a limited change in parole
procedures designed to relieve the state "from the costly and time
consuming responsibility of scheduling parole hearings" for prisoners

                    17
who have a low probability of being released. 

(citation omitted). The lesson of Morales is that the state may
adopt such a change unless a prisoner can "show`with assurance' . . .
that he would have received parole under the old system." Johnson v.
Gomez, 

, 968 (9th Cir. 1996) (citations omitted). At best,
Roller can contend that he will miss an "opportunity to take advantage
of provisions for early release," and under Morales that is simply
insufficient to establish an ex post facto 

n.3 (emphasis in original).

IV.

We hold that the Prison Litigation Reform Act's filing fee and pay-
ment of costs requirements are constitutional. We affirm the district
court's judgment that the application of section 24-21-645 to plaintiff
does not constitute a violation of the Ex Post Facto Clause.

AFFIRMED

HALL, Circuit Judge, concurring in part and dissenting in part:

I agree that the PLRA withstands the constitutional attacks levelled
upon it by Roller, and I therefore concur in Part II of Chief Judge
Wilkinson's opinion. I do not believe, however, that Morales dictates
the result reached by the majority in Part III of its opinion, and, there-
fore, I dissent from the affirmance of the district court's judgment that
South Carolina's amendment to its parole statutes is constitutional.

A

In concluding that the California statute did not violate the Ex Post
Facto clause because it "create[d] only the most speculative and atten-
uated possibility of producing the prohibited effect of increasing the
measure of punishment for covered crimes," the Supreme Court in
Morales emphasized two features of the statute: (1) the amended pro-
cedure "applies only to a class of prisoners for whom the likelihood
of release on parole is quite remote," and (2) the Parole Board's
authority was carefully tailored to achieve the amendment's purpose
of eliminating hearings "for prisoners who have no chance of being

                     18
released." 

-04. Unlike the situation pres-
ented in Hill v. Jackson, 

(4th Cir. 1995), where we
applied the Morales analysis to uphold a Virginia statute that changed
the frequency of parole-eligibility hearings, the South Carolina
amendments do not "bear[ ] a strong resemblance to the California
statute sustained in Morales," ante at 14. Comparison of the three
statutes (California, Virginia,1 and South Carolina) manifests that the
degree of change effected by the South Carolina law is one that is of
"sufficient moment to transgress the constitutional prohibition."

(quoting Beazell v. Ohio, 

,
171 (1925)).

B

The statute in Morales applies only to prisoners who had been
"convicted of more than one offense involving the taking of a life."
The Virginia policy considered in Hill applies "only to a narrow class
of inmates," estimated by the Parole Board to comprise some 3% of
the prison population, who are serving (1) a life sentence, or (2) a sen-
tence or sentences totalling at least 70 years for at least one violent
offense and for which there are at least 10 years left to be served
before mandatory release. 

. Clearly, then, these
two statutes affect only the worst of the worst.

The South Carolina statute, on the other hand, applies to all
inmates convicted of a "violent crime," as that term is defined in S.C.
Code Ann. § 16-1-60 (Law. Co-op. Supp. 1995)."Violent crime"
includes far more offenses than those covered by the deferral rules in
California and Virginia. For instance, § 16-1-60 includes crimes for
which only a 1-10 year sentence is prescribed. See, e.g., S.C. Code
§ 44-53-370(e)(1)(a) (Law. Co-op. Supp. 1995) (sale, cultivation,
possession, etc., of 10-100 pounds of marijuana). We are simply left
to guess about how many South Carolina inmates are affected by this
provision because the district court made no findings in this regard,
and the State makes no effort on this appeal to point out any legisla-
tive findings or other authority that would suggest that the amendment
_________________________________________________________________
1 Although Hill dealt with a Parole Board policy that was appreciably
narrower in scope than the statute permitted, I will use the term "statute"
in referring to the law involved in that case.

                    19
in question "applies only to a class of prisoners for whom the likeli-
hood of release on parole is quite remote." 

. A comparison of the facial differences among the three groups
affected -- multiple murderers in California, those with at least a 70-
year sentence in Virginia, and certain drug traffickers, burglars, sex
offenders and others in South Carolina serving as little as 1-10 years
-- demonstrates that the South Carolina law casts a much wider net.
Contrary to the majority's assertion, then, the South Carolina statute
does not, at least with respect to the class of persons affected, "bear[ ]
a strong resemblance to the California statute sustained in Morales."
Ante at 14.

C

I turn, then, to the second feature of the California statute noted in
Morales, the "careful[ ] tailor[ing]" of the Board's authority to relieve
it of the "costly and time-consuming responsibility of scheduling
parole hearings for prisoners who have no chance of being 

(quoting In re Jackson, 

, 473, 

, 764, 

, 105 (1985)). In all three statutes,
the date of the initial parole hearing is unaffected and the substantive
standards for eligibility remain the same. But a detailed comparison
of the procedures regarding deferrals in California and Virginia on the
one hand, and South Carolina on the other, illustrates starkly critical
differences.

First, the "default requirement" in Virginia and California is annual
review. In California, the Board must conclude that"it is not reason-
able to expect that parole would be granted at a hearing during the
following years," and it must state the bases for such finding.

. Similarly, deferrals in Virginia "are not
automatic." 

. In other words, every inmate in these
states will receive annual hearings unless the Parole Board affirma-
tively decides that deferral is warranted on the basis of the evidence
considered at the last hearing. In South Carolina, however, the default
requirement for the affected class of inmates is two years, and there
is no provision requiring a finding that deferral is warranted.

Second, deferral decisions in Virginia are subject to appeal at any
time during the deferral period. The Supreme Court noted that while

                     20
the California law was unclear, "the reliability of the Board's [defer-
ral] determination may also be enhanced by the possibility of an
administrative appeal." 

. Although the
majority finds no express statutory restriction on the South Carolina
Board's authority to schedule expedited hearings, ante at 15, there is
neither evidence in the record nor assertion by the State that such
authority exists. In any event, there is clearly no provision for an
appeal from the Board's refusal to schedule an expedited hearing. In
other words, while the Board might be able to get around the two-year
rule, an inmate cannot challenge a decision not to expedite a hearing.

Third, the Supreme Court noted in Morales that the possibility of
immediate release after a finding of suitability for parole is largely
"theoretica[l]" and that "in many cases, the prisoner's parole release
date comes at least several years after a finding of suitability." 

see also In re 

(noting that several
inmate petitions received by that court involved periods between the
date of the suitability finding and the proposed release date that varied
from three and a half years to nineteen years). 2 There is no reason to
believe that anything approaching this situation obtains in South Car-
olina. See S.C. Code Ann. § 24-21-645 (providing for a 90-day "pro-
visional parole" period prior to release).

D

The amendment to the South Carolina parole statute made another
change that has no counterpart in the statutes under consideration in
Morales and Hill. When Roller was sentenced in 1983, the parole
statute in effect provided for parole upon authorization by a majority
of the Board. 1981 S.C. Acts No. 100 § 13. Three years after he was
sentenced, the provision was amended to require that at least two-
thirds of the Board's members were needed to authorize the parole of
persons convicted of a "violent crime," 1986 S.C. Acts No. 462 § 31,
_________________________________________________________________
2 The hearings at issue in Morales involved the suitability for parole.
The actual release date, however, is determined by a matrix of facts relat-
ing to the crime and victim. According to the State of California, even
had Morales been found suitable for parole at the initial hearing, he could
not have been released until approximately 2001. See Petitioner's brief,
U.S.S.Ct., 

*22 n.8.

                    21
and that two-thirds requirement remains in effect today, codified in
§ 24-21-645. The practical effect of this is that the inmate must now
garner the votes of five Parole Board members rather than a simple
majority of four. The majority considers this change apart from the
other retrospective changes in the statute and declares that any ex post
facto challenge is foreclosed by Dobbert v. Florida, 

(1977). I disagree.

As a preliminary matter, I believe that the two-thirds requirement
must be considered together with the other changes to the parole stat-
ute. See 

at 294 ("We must compare the two statutory
procedures in toto to determine if the new may be fairly characterized
as more onerous."). If, as I argue below, the new two-thirds require-
ment makes parole tougher to attain, then this factor must be exam-
ined in conjunction with the amendment's decrease in the frequency
of the hearings to determine whether the overall changes to the statute
violate the Ex Post Facto Clause.

The majority's view that the two-thirds requirement might actually
inure to the inmate's advantage flows from neither Dobbert nor com-
mon sense. See ante at 16 ("[P]arole board members might be more
likely to vote for granting parole under the two-thirds rule, knowing
that any favorable decision must be concurred in by a greater number
of their colleagues."). The footnote in Dobbert to which the majority
refers, ante at 16 (citing 

n.7), was merely the Supreme
Court's response to the petitioner's argument that the discrete portion
of the amendment under consideration -- permitting trial judge
review of a jury recommendation of life imprisonment in a capital
case -- was more onerous than the prior statute that left the sentenc-
ing decision (death or life imprisonment) up to the jury. The Court
went on to find that the overall changes, of which trial judge review
was only part, "afford[ ] significantly more safeguards to the defen-
dant than did the old [statute]," 

so much so that the Court
characterized the overall changes as "ameliorative." 

No one
would argue that inmates affected by the South Carolina amendments
are now in a better position than they were before the statutory
changes.

The majority also notes that Morales compels upholding the two-
thirds requirement because to do otherwise would amount to the judi-

                    22
cial "micromanagement" that the Court cautioned against. Ante at 16.
Morales does no such thing. The California statute involved an
exceedingly speculative possibility that the punishment of the affected
inmates would be increased: The statute applies only to multiple mur-
derers, presumably a small fraction of the inmate population; the
Board has to affirmatively decide that a hearing should be deferred
and to explain why; the inmate might be able to appeal the deferral
decision, and the Board could, of its own volition, advance a hearing
date where a change in circumstances warranted; and, significantly,
under California's system, the determination of parole suitability
often precedes the actual release date by several years. South Caroli-
na's amendments, on the other hand, affect persons convicted of rela-
tively minor crimes; mandate automatic deferrals, with no provision
for an administrative appeal; increase the percentage of the Board that
must vote to grant parole. In addition, there is no indication that a
grant of parole is not ordinarily followed promptly by actual release.
If Morales is our guide, the South Carolina statute increases the pun-
ishment by decreasing the likelihood of release on parole to a degree
that offends the Ex Post Facto Clause.

                    23