delivered the opinion of the Court.
Title 38 U. S. C. § 3404(c) limits to $10 the fee that may be paid an attorney or agent who represents a veteran seeking benefits for service-connected death or disability. The United States District Court for the Northern District of California held that this limit violates the Due Process Clause of the Fifth Amendment, and the First Amendment, because it denies veterans or their survivors the opportunity to retain counsel of their choice in pursuing their claims. We noted probable jurisdiction of the Government’s appeal, 469 U. S. 1085 (1984), and we now reverse.
I — I
Congress has by statute established an administrative system for granting service-connected death or disability benefits to veterans. See 38 U. S. C. § 301 et seq. The amount of the benefit award is not based upon need, but upon service connection — that is, whether the disability is causally related to an injury sustained in the service — and the degree of incapacity caused by the disability. A detailed system has been established by statute and Veterans’ Administration (VA) regulation for determining a veteran’s entitlement, with final authority resting with an administrative body known as the Board of Veterans’ Appeals (BVA). Judicial review of VA decisions is precluded by statute. 38 U. S. C. § 211(a); Johnson v. Robison, 415 U. S. 361 (1974). The controversy in this case centers on the opportunity for a benefit applicant *308or recipient to obtain legal counsel to aid in the presentation of his claim to the VA. Section 3404(c) of Title 38 provides:
“The Administrator shall determine and pay fees to agents or attorneys recognized under this section in allowed claims for monetary benefits under laws administered by the Veterans’ Administration. Such fees—
“(2) shall not exceed $10 with respect to any one claim . . .
Section 3405 provides criminal penalties for any person who charges fees in excess of the limitation of § 3404.
Appellees here are two veterans’ organizations, three individual veterans, and a veteran’s widow.1 The two veterans’ organizations are the National Association of Radiation Survivors, an organization principally concerned with obtaining compensation for its members for injuries resulting from atomic bomb tests, and Swords to Plowshares Veterans Rights Organization, an organization particularly devoted to the concerns of Vietnam veterans. The complaint contains no further allegation with respect to the numbers of members in either organization who are veteran claimants. Appellees did not seek class certification.
Appellees contended in the District Court that the fee limitation provision of § 3404 denied them any realistic opportunity to obtain legal representation in presenting their claims to the VA and hence violated their rights under the Due Process Clause of the Fifth Amendment and under the First Amendment. The District Court agreed with the appellees on both of these grounds, and entered a nationwide “preliminary injunction” barring appellants from enforcing the fee limitation. 589 F. Supp. 1302 (1984). To understand fully the posture in which the case reaches us it is necessary to discuss the administrative scheme in some detail.
*309Congress began providing veterans pensions in early 1789, and after every conflict in which the Nation has been involved Congress has, in the words of Abraham Lincoln, “provided for him who has borne the battle, and his widow and his orphan.” The VA was created by Congress in 1930, and since that time has been responsible for administering the congressional program for veterans’ benefits. In 1978, the year covered by the report of the Legal Services Corporation to Congress that was introduced into evidence in the District Court, approximately 800,000 claims for service-connected disability or death and pensions were decided by the 58 regional offices of the VA. Slightly more than half of these were claims for service-connected disability or death, and the remainder were pension claims. Of the 800,000 total claims in 1978, more than 400,000 were allowed, and some 379,000 were denied. Sixty-six thousand of these denials were contested at the regional level; about a quarter of these contests were dropped, 15% prevailed on reconsideration at the local level, and the remaining 36,000 were appealed to the BVA. At that level some 4,500, or 12%, prevailed, and another 13% won a remand for further proceedings. Although these figures are from 1978, the statistics in evidence indicate that the figures remain fairly constant from year to year.
As might be expected in a system which processes such a large number of claims each year, the process prescribed by Congress for obtaining disability benefits does not contemplate the adversary mode of dispute resolution utilized by courts in this country. It is commenced by the submission of a claim form to the local veterans agency, which form is provided by the VA either upon request or upon receipt of notice of the death of a veteran. Upon application a claim generally is first reviewed by a three-person “rating board” of the VA regional office — consisting of a medical specialist, a legal specialist, and an “occupational specialist.” A claimant is “entitled to a hearing at any time on any issue involved in a claim . . . .” 38 CFR §3.103(c) (1984). Proceedings in front of the rating board “are ex parte in nature,” §3.103(a); no *310Government official appears in opposition. The principal issues are the extent of the claimant’s disability and whether it is service connected. The board is required by regulation “to assist a claimant in developing the facts pertinent to his claim,” § 3.103(a), and to consider any evidence offered by the claimant. See §3.103(b). In deciding the claim the board generally will request the applicant’s Armed Service and medical records, and will order a medical examination by a VA hospital. Moreover, the board is directed by regulation to resolve all reasonable doubts in favor of the claimant. §3.102.2
After reviewing the evidence the board renders a decision either denying the claim or assigning a disability “rating” pursuant to detailed regulations developed for assessing various disabilities. Money benefits are calculated based on the rating. The claimant is notified of the board’s decision and its reasons, and the claimant may then initiate an appeal by *311filing a “notice of disagreement” with the local agency. If the local agency adheres to its original decision it must then provide the claimant with a “statement of the case” — a written description of the facts and applicable law upon which the board based its determination — so that the claimant may adequately present his appeal to the BVA. Hearings in front of the BVA are subject to the same rules as local agency hearings — they are ex parte, there is no formal questioning or cross-examination, and no formal rules of evidence apply. 38 CFR §19.157 (1984). The BVA’s decision is not subject to judicial review. 38 U. S. C. § 211(a).3
The process is designed to function throughout with a high degree of informality and solicitude for the claimant. There is no statute of limitations, and a denial of benefits has no formal res judicata effect; a claimant may resubmit as long as he presents new facts not previously forwarded. See 38 CFR §§3.104, 3.105 (1984). Although there are time limits for submitting a notice of disagreement and although a claimant may prejudice his opportunity to challenge factual or legal decisions by failing to challenge them in that notice, the time limit is quite liberal — up to one year — and the VA boards are instructed to read any submission in the light most favorable to the claimant. See 38 CFR §§19.129, 19.124, 19.121 (1984). Perhaps more importantly for present purposes, however, various veterans’ organizations across the country make available trained service agents, free of charge, to assist claimants in developing and presenting their claims. These service representatives are contemplated by the VA statute, 38 U. S. C. § 3402, and they are recognized as an important part of the administrative scheme. Appellees’ counsel agreed at argument that a representative is available for *312any claimant who requests one, regardless of the claimant’s affiliation with any particular veterans’ group.4
In support of their claim that the present statutory and administrative scheme violates the Constitution, appellees submitted affidavits and declarations of 16 rejected claimants or recipients and 24 practicing attorneys, depositions of several VA employees, and various exhibits. The District Court held a hearing and then issued a 52-page opinion and order granting the requested “preliminary injunction.”5
With respect to the merits of appellees’ due process claim, the District Court first determined that recipients of service-connected death and disability benefits possess “property” interests protected by the Due Process Clause, see Mathews v. Eldridge, 424 U. S. 319 (1976) (recipients of Social Security benefits possess a protected “property” interest), and also held that applicants for such benefits possess such an interest. Although noting that this Court has never ruled on the latter question, the court relied on several opinions of the Court of Appeals for the Ninth Circuit holding, with respect to similar Government benefits, that applicants possess such an interest. See, e. g., Ressler v. Pierce, 692 F. 2d 1212, 1214-1216 (1982) (applicants for federal rent subsidies).
The court then held that appellees had a strong likelihood of showing that the administrative scheme violated the due process rights of those entitled to benefits. In holding that the process described above was “fundamentally unfair,” the court relied on the analysis developed by this Court in *313Mathews v. Eldridge, swpra, in which we stated the factors that must be weighed in determining what process is due an individual subject to a deprivation:
“First, the private interest that will be affected by the official action; second, the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards; and finally, the Government’s interest, including the function involved and the fiscal and administrative burdens that the additional or substitute procedural requirement would entail.” 424 U. S., at 335.
In applying this test the District Court relied heavily on appellees’ evidence; it noted that the veterans’ interest in receiving benefits was significant in that many recipients are disabled, and totally or primarily dependent on benefits for their support. 589 F. Supp., at 1315. With respect to the likelihood of error under the present system, and the value of the additional safeguard of legal representation, it first noted that some of the appellees had been represented by service agents and had been dissatisfied with their representation, and had sought and failed to obtain legal.counsel due solely to the fee limitation. The court found that absent expert legal counsel claimants ran a significant risk of forfeiting their rights, because of the highly complex issues involved in some cases. VA processes, the court reasoned, allow claimants to waive points of disagreement on appeal, or to waive appeal altogether by failing to file the notice of disagreement; in addition, claimants simply are not equipped to engage in the factual or legal development necessary in some cases, or to spot errors made by the administrative boards. Id., at 1319-1321.
With respect to whether the present process alleviated these problems, the court found that “neither the VA officials themselves nor the service organizations are providing the full array of services that paid attorneys might make avail*314able to claimants.” Id., at 1320. Even assuming that all VA personnel were willing to go out of their way for each claimant, a point which the court would not fully accept,6 the court found that in any event the VA does not have the resources to permit the substantial investments of time that are necessary. The VA does not seek independent testimony that might establish service connection, or independent medical examinations with respect to disability.
In reaching its conclusions the court relied heavily on the problems presented by what it described as “complex cases” — a class of cases also focused on in the depositions. Though never expressly defined by the District Court, these cases apparently include those in which a disability is slow developing and therefore difficult to find service connected, such as the claims associated with exposure to radiation or harmful chemicals, as well as other cases identified by the deponents as involving difficult matters of medical judgment. Nowhere in the opinion of the District Court is there any estimate of what percentage of the annual VA caseload of 800,000 these cases comprise, nor is there any more precise description of the class. There is no question but what the 3 named plaintiffs and the plaintiff veteran’s widow asserted such claims, and in addition there are declarations in the record from 12 other claimants who were asserting such claims. The evidence contained in the record, however, suggests that the sum total of such claims is extremely small; in 1982, for example, roughly 2% of the BVA caseload consisted of “agent orange” or “radiation” claims, and what evidence *315there is suggests that the percentage of such claims in the' regional offices was even less — perhaps as little as 3 in 1,000.
With respect to the service representatives, the court again found the representation unsatisfactory. Although admitting that this was not due to any “lack of dedication,” the court found that a heavy caseload and the lack of legal training combined to prevent service representatives from adequately researching a claim. Facts are not developed, and “it is standard practice for service organization representatives to submit merely a one to two page handwritten brief.” Id., at 1322.
Based on the inability of the VA and service organizations to provide the full range of services that a retained attorney might, the court concluded that appellees had demonstrated a “high risk of erroneous deprivation” from the process as administered. Ibid. The court then found that the Government had “failed to demonstrate that it would suffer any harm if the statutory fee limitation . . . were lifted.” Id., at 1323. The only Government interest suggested was the “paternalistic” assertion that the fee limitation is necessary to ensure that claimants do not turn substantial portions of their benefits over to unscrupulous lawyers. The court suggested that there were “less drastic means” to confront this problem.
Finally, the court agreed with appellees that there was a substantial likelihood that the fee limitation also violates the First Amendment. The court relied on this Court’s decisions in Mine Workers v. Illinois Bar Assn., 389 U. S. 217 (1967), and Railroad Trainmen v. Virginia ex rel. Virginia State Bar, 377 U. S. 1 (1964), as establishing “the principle that the First Amendment rights to petition, association and speech protect efforts by organizations and individuals to obtain effective legal representation of their constituents or themselves.” 589 F. Supp., at 1324. This right to “adequate legal representation” or “meaningful access to courts,” the court found, was infringed by the fee limitation — again *316without substantial justification by the Government. Id., at 1325-1326.
After reiterating the Government’s failure of proof with respect to the likely harms arising from doing away with the fee limitation, the court entered a “preliminary injunction” enjoining the Government appellants from “enforcing or attempting to enforce in any way the provisions of 38 U. S. C. §§3404-3405 . . . .” Id., at 1329. The injunction was not limited to the particular plaintiffs, nor was it limited to claims processed in the District of Northern California, where the court sits.
II
Before proceeding to the merits we must deal with a significant question as to our jurisdiction, one not raised by appellees in this Court. This appeal was taken under 28 U. S. C. § 1252, which grants this Court jurisdiction “from an interlocutory or final judgment, decree or order of any court of the United States . . . holding an Act of Congress unconstitutional in any civil action ... to which the United States or any of its agencies, or any officer or employee thereof, as such officer or employee, is a party.” We have here an interlocutory decree in a civil action to which an officer of the United States is a party, and the only question is whether the District Court’s decision “holds” an Act of Congress unconstitutional. The problem, of course, is that given that the court’s opinion and order are cast in terms of a “preliminary injunction” the court only states that there is a “high likelihood of success” on the merits of appellees’ claims, and does not specifically state that the fee limitation provision is unconstitutional.
We do not write on a clean slate. In McLucas v. DeChamplain, 421 U. S. 21 (1975), this Court similarly entertained an appeal from an order that granted a preliminary injunction and in the process held an Act of Congress unconstitutional. In holding that we had jurisdiction under § 1252 we noted that that section constitutes an “exception” to “the *317policy ... of minimizing the mandatory docket of this Court,” and we went on to state:
“It might be argued that, in deciding to issue the preliminary injunction, the District Court made only an interlocutory determination of appellee’s probability of success on the merits and did not finally ‘hold’ the article unconstitutional. By its terms, however, § 1252 applies to interlocutory as well as final judgments, decrees, and orders, and this Court previously has found the section properly invoked when the court below has made only an interlocutory determination of unconstitutionality, at least if, as here, that determination forms the necessary predicate to the grant or denial of preliminary equitable relief.” Id., at 30.
We think this case is controlled by McLucas. It is true that in McLucas the District Court actually stated its holding that the statute was unconstitutional, whereas here the court’s statements are less direct. But that is merely a semantic difference in this case; inasmuch as any conclusions reached at the preliminary injunction stage are subject to revision, University of Texas v. Camenisch, 451 U. S. 390, 395 (1981), it should make little difference whether the court stated conclusively that a statute was unconstitutional, or merely said it was likely, so long as the injunction granted enjoined the statute’s operation. This Court’s appellate jurisdiction does not turn on such semantic niceties. See also California v. Grace Brethren Church, 457 U. S. 393, 405 (1982) (“§ 1252 provides jurisdiction even though the lower court did not expressly declare a federal statute unconstitutional . . .”).
Indeed, we note that the problem raised by the statute’s use of the word “holding” may in any event be a bit of a red herring. In its original form § 1252 provided this Court with appellate jurisdiction over decisions “against the constitutionality of any Act of Congress,” see Act of Aug. 24, 1937, *318ch. 754, § 2, 50 Stat. 752;7 although this language was changed when the provision was codified in 1948, so that § 1252 now grants jurisdiction from a decision “holding any Act of Congress unconstitutional,” this change was effected without substantive comment, and absent such comment it is generally held that a change during codification is not intended to alter the statute’s scope. See Muniz v. Hoffman, 422 U. S. 454, 467-474 (1975). Any fair reading of the decision at issue would conclude that it is “against the constitutionality” of §3404, and we are loath to read an unheralded change in phraseology to divest us of jurisdiction here.
Finally, acceptance of appellate jurisdiction in this case is in accord with the purpose of the statutory grant. Last Term, in Heckler v. Edwards, 465 U. S. 870 (1984), we discussed § 1252’s legislative history. We noted that in enacting § 1252 Congress sought to identify a category of important decisions adverse to the constitutionality of an Act of Congress — which decisions, because the United States or its agent was a party, had implications beyond the controversy then before the court — and to provide an expeditious means for ensuring certainty and uniformity in the enforcement of such an Act by establishing direct review over such decisions in this Court. Id., at 879-883. Edwards teaches that the decisions Congress targeted for appeal under § 1252 were those which involved the exercise of judicial power to impair the enforcement of an Act of Congress on constitutional grounds, and that it was the constitutional question that Congress wished this Court to decide. As we pointed out in McLucas, *319§1252 contemplates that this impairment can arise from interlocutory decrees, just as the original statute provided for appeal from decisions in “any proceedings.” Cf. Goldstein v. Cox, 396 U. S. 471, 476 (1970) (28 U. S. C. § 1253 authorizes direct appeals from preliminary injunctions issued by three-judge courts). A single district judge’s interlocutory decision on constitutional grounds that an Act of Congress should not be enforced frustrates the will of Congress in the short run just as surely as a final decision to that effect. By § 1252 Congress gave the Government the right of immediate appeal to this Court in such a situation so that only those district court injunctions which had been reviewed and upheld by this Court would continue to have such an effect. Cf. Edwards, swpra. The injunction at issue here creates precisely the problem to which § 1252 was addressed, inasmuch as it enjoins the operation of the fee limitation on constitutional grounds, across the country and under all circumstances. Thus, whether or not the injunction here is framed as a “holding” of unconstitutionality we believe we have jurisdiction under § 1252.
HH I — i ► — I
Judging the constitutionality of an Act of Congress is properly considered “ ‘the gravest and most delicate duty that this Court is called upon to perform,’” Rostker v. Goldberg, 453 U. S. 57, 64 (1981) (quoting Blodgett v. Holden, 275 U. S. 142, 148 (1927) (Holmes, J.)), and we begin our analysis here with no less deference than we customarily must pay to the duly enacted and carefully considered decision of a coequal and representative branch of our Government. Indeed one might think, if anything, that more deference is called for here; the statute in question for all relevant purposes has been on the books for over 120 years. Cf. McCulloch v. Maryland, 4 Wheat. 316, 401-402 (1819). This deference to congressional judgment must be afforded even though the *320claim is that a statute Congress has enacted effects a denial of the procedural due process guaranteed by the Fifth Amendment. Schweiker v. McClure, 456 U. S. 188 (1982); Mathews v. Eldridge, 424 U. S., at 349. We think that the District Court went seriously awry in assessing the constitutionality of §3404.
Appellees’ first claim, accepted by the District Court, is that the statutory fee limitation, as it bears on the administrative scheme in operation, deprives a rejected claimant or recipient of “life, liberty or property, without due process of law,” U. S. Const., Arndt. 5, by depriving him of representation by expert legal counsel.8 Our decisions establish that “due process” is a flexible concept — that the processes required by the Clause with respect to the termination of a protected interest will vary depending upon the importance attached to the interest and the particular circumstances under which the deprivation may occur. See Mathews, supra, at 334; Morrissey v. Brewer, 408 U. S. 471, 481 (1972). In defining the process necessary to ensure “fundamental fairness” we have recognized that the Clause does not require that “the procedures used to guard against an erroneous deprivation ... be so comprehensive as to preclude any possibility of error,” Mackey v. Montrym, 443 U. S. 1, 13 (1979), and in addition we have emphasized that the marginal gains from affording an additional procedural safeguard often may be *321outweighed by the societal cost of providing such a safeguard. See Mathews, 424 U. S., at 348.9
These general principles are reflected in the test set out in Mathews, which test the District Court purported to follow, and which requires a court to consider the private interest that will be affected by the official action, the risk of an erroneous deprivation of such interest through the procedures used, the probable value of additional or substitute procedural safeguards, and the government’s interest in adhering to the existing system. Id., at 335. In applying this test we must keep in mind, in addition to the deference owed to Congress, the fact that the very nature of the due process inquiry indicates that the fundamental fairness of a particular procedure does not turn on the result obtained in any individual case; rather, “procedural due process rules are shaped by the risk of error inherent in the truth-finding process as applied to the generality of cases, not the rare exceptions.” Id., at 344; see also Parham v. J. R., 442 U. S. 584, 612-613 (1979).
The Government interest, which has been articulated in congressional debates since the fee limitation was first enacted in 1862 during the Civil War, has been this: that the system for administering benefits should be managed in a sufficiently informal way that there should be no need for the employment of an attorney to obtain benefits to which a claimant was entitled, so that the claimant would receive the entirety of the award without having to divide it with a lawyer. See United States v. Hall, 98 U. S. 343, 352-355 (1879). This purpose is reinforced by a similar absolute prohibition on compensation of any service organization repre*322sentative. 38 U. S. C. 3402(b)(1). While Congress has recently considered proposals to modify the fee limitation in some respects, a Senate Committee Report in 1982 highlighted that body’s concern that “any changes relating to attorneys’ fees be made carefully so as not to induce unnecessary retention of attorneys by VA claimants and not to disrupt unnecessarily the very effective network of non-attorney resources that has evolved in the absence of significant attorney involvement in VA claims matters.” S. Rep. No. 97-466, p. 49 (1982). Although this same Report professed the Senate’s belief that the original stated interest in protecting veterans from unscrupulous lawyers was “no longer tenable,” the Senate nevertheless concluded that the fee limitation should with a limited exception remain in effect, in order to “protect claimants’ benefits” from being unnecessarily diverted to lawyers.10
In the face of this congressional commitment to the fee limitation for more than a century, the District Court had only this to say with respect to the governmental interest:
“The government has neither argued nor shown that lifting the fee limit would harm the government in any way, *323except as the paternalistic protector of claimants’ supposed best interests. To the extent the paternalistic role is valid, there are less drastic means available to ensure that attorneys’ fees do not deplete veterans’ death or disability benefits.” 589 F. Supp., at 1323.
It is not for the District Court or any other federal court to invalidate a federal statute by so cavalierly dismissing a long-asserted congressional purpose. If “paternalism” is an insignificant Government interest, then Congress first went astray in 1792, when by its Act of March 23 of that year it prohibited the “sale, transfer or mortgage ... of the pension . . . [of a] soldier . . . before the same shall become due.” Ch. 11, § 6, 1 Stat. 245. Acts of Congress long on the books, such as the Fair Labor Standards Act, might similarly be described as “paternalistic”; indeed, this Court once opined that “[statutes of the nature of that under review, limiting the hours in which grown and intelligent men may labor to earn their living, are mere meddlesome interferences with the rights of the individual . . . .” Lochner v. New York, 198 U. S. 45, 61 (1905). That day is fortunately long gone, and with it the condemnation of rational paternalism as a legitimate legislative goal.
There can be little doubt that invalidation of the fee limitation would seriously frustrate the oft-repeated congressional purpose for enacting it. Attorneys would be freely employable by claimants to veterans’ benefits, and the claimant would as a result end up paying part of the award, or its equivalent, to an attorney. But this would not be the only consequence of striking down the fee limitation that would be deleterious to the congressional plan.
A necessary concomitant of Congress’ desire that a veteran not need a representative to assist him in making his claim was that the system should be as informal and nonadversarial as possible. This is not to say that complicated factual inquiries may be rendered simple by the expedient of informality, but surely Congress desired that the proceedings be as *324informal and nonadversarial as possible.11 The regular introduction of lawyers into the proceedings would be quite unlikely to further this goal. Describing the prospective impact of lawyers in probation revocation proceedings, we said in Gagnon v. Scarpelli, 411 U. S. 778, 787-788 (1973):
“The introduction of counsel into a revocation proceeding will alter significantly the nature of the proceeding. If counsel is provided for the probationer or parolee, the State in turn will normally provide its own counsel; lawyers, by training and disposition, are advocates and bound by professional duty to present all available evidence and arguments in support of their clients’ positions and to contest with vigor all adverse evidence and views. The role of the hearing body itself. . . may become more akin to that of a judge at a trial, and less attuned to the rehabilitative needs of the individual .... Certainly, *325the decisionmaking process will be prolonged, and the financial cost to the State — for appointed counsel, ... a longer record, and the possibility of judicial review — will not be insubstantial.”
We similarly noted in Wolff v. McDonnell, 418 U. S. 539, 570 (1974), that the use of counsel in prison disciplinary proceedings would “inevitably give the proceedings a more adversary cast. . . .”
Knowledgeable and thoughtful observers have made the same point in other language:
“To be sure, counsel can often perform useful functions even in welfare cases or other instances of mass justice; they may bring out facts ignored by or unknown to the authorities, or help to work out satisfactory compromises. But this is only one side of the coin. Under our adversary system the role of counsel is not to make sure the truth is ascertained but to advance his client’s cause by any ethical means. Within the limits of professional propriety, causing delay and sowing confusion not only are his right but may be his duty. The appearance of counsel for the citizen is likely to lead the government to provide one — or at least to cause the government’s representative to act like one. The result may be to turn what might have been a short conference leading to an amicable result into a protracted controversy.
“These problems concerning counsel and confrontation inevitably bring up the question whether we would not do better to abandon the adversary system in certain areas of mass justice. . . . While such an experiment would be a sharp break with our tradition of adversary process, that tradition, which has come under serious general challenge from a thoughtful and distinguished judge, was not formulated for a situation in which many thousands of hearings must be provided each month.” *326Friendly, “Some Kind of Hearing,” 123 U. Pa. L. Rev. 1267, 1287-1290 (1975).
Thus, even apart from the frustration of Congress’ principal goal of wanting the veteran to get the entirety of the award, the destruction of the fee limitation would bid fair to complicate a proceeding which Congress wished to keep as simple as possible. It is scarcely open to doubt that if claimants were permitted to retain compensated attorneys the day might come when it could be said that an attorney might indeed be necessary to present a claim properly in a system rendered more adversary and more complex by the very presence of lawyer representation. It is only a small step beyond that to the situation in which the claimant who has a factually simple and obviously deserving claim may nonetheless feel impelled to retain an attorney simply because so many other claimants retain attorneys. And this additional complexity will undoubtedly engender greater administrative costs, with the end result being that less Government money reaches its intended beneficiaries.
We accordingly conclude that under the Mathews v. Eldridge analysis great weight must be accorded to the Government interest at stake here. The flexibility of our approach in due process cases is intended in part to allow room for other forms of dispute resolution; with respect to the individual interests at stake here, legislatures are to be allowed considerable leeway to formulate such processes without being forced to conform to a rigid constitutional code of procedural necessities. See Parham v. J. R., 442 U. S., at 608, n. 16. It would take an extraordinarily strong showing of probability of error under the present system — and the probability that the presence of attorneys would sharply diminish that possibility — to warrant a holding that the fee limitation denies claimants due process of law. We have no hesitation in deciding that no such showing was made out on the record before the District Court.
*327As indicated by the statistics set out earlier in this opinion, more than half of the 800,000 claims processed annually by the VA result in benefit awards at the regional level. An additional 10,000 claims succeed on request for reconsideration at the regional level, and of those that do not, 36,000 are appealed to the BVA. Of these, approximately 16% succeed before the BVA. It is simply not possible to determine on this record whether any of the claims of the named plaintiffs, or of other declarants who are not parties to the action, were wrongfully rejected at the regional level or by the BVA, nor is it possible to quantify the “erroneous deprivations” among the general class of rejected claimants. If one regards the decision of the BVA as the “correct” result in every case, it follows that the regional determination against the claimant is “wrong” in the 16% of the cases that are reversed by the Board.
Passing the problems with quantifying the likelihood of an erroneous deprivation, however, under Mathews we must also ask what value the proposed additional procedure may have in reducing such error. In this case we are fortunate to have statistics that bear directly on this question, which statistics were addressed by the District Court. These unchallenged statistics chronicle the success rates before the BVA depending on the type of representation of the claimant, and are summarized in the following figures taken from the record. App. 568.
Ultimate Success Rates Before the Board of Veterans’ Appeals by Mode of Representation
American Legion . 16.2%
American Red Cross . 16.8%
Disabled American Veterans . 16.6%
Veterans of Foreign Wars . 16.7%
Other nonattorney . 15.8%
No representation . 15.2%
Attorney/Agent . 18.3%
*328The District Court opined that these statistics were not helpful, because in its view lawyers were retained so infrequently that no body of lawyers with an expertise in VA practice had developed, and lawyers who represented veterans regularly might do better than lawyers who represented them only pro bono on a sporadic basis. The District Court felt that a more reliable index of the effect lawyers would have on the proceedings was a statistical study showing success of various representatives in appeals to discharge review boards in the uniformed services — statistics that showed a significantly higher success rate for those claimants represented by lawyers as compared to those claimants not so represented.
We think the District Court’s analysis of this issue totally unconvincing, and quite lacking in the deference which ought to be shown by any federal court in evaluating the constitutionality of an Act of Congress. We have the most serious doubt whether a competent lawyer taking a veteran’s case on a pro bono basis would give less than his best effort, and we see no reason why experience in developing facts as to causation in the numerous other areas of the law where it is relevant would not be readily transferable to proceedings before the VA. Nor do we think that lawyers’ success rates in proceedings before military boards to upgrade discharges — proceedings which are not even conducted before the VA, but before military boards of the uniformed services — are to be preferred to the BVA statistics which show reliable success by mode of representation in the very type of proceeding to which the litigation is devoted.
The District Court also concluded, apparently independently of its ill-founded analysis of the claim statistics, (1) that the VA processes are procedurally, factually, and legally complex, and (2) that the VA system presently does not work as designed, particularly in terms of the representation afforded by VA personnel and service representatives, and that these representatives are “unable to perform all of the services which might be performed by a claimant’s own *329paid attorney.” 589 F. Supp., at 1322. Unfortunately the court’s findings on “complexity” are based almost entirely on a description of the plan for administering benefits in the abstract, together with references to “complex” cases involving exposure to radiation or agent orange, or post-traumatic stress syndrome. The court did not attempt to state even approximately how often procedural or substantive complexities arise in the run-of-the-mine case, or even in the unusual case. The VA procedures cited by the court do permit a claimant to prejudice his rights by failing to respond in a timely manner to an agency notice of denial of an initial claim, but despite this possibility there is nothing in the District Court’s opinion indicating that these procedural requirements have led to an unintended forfeiture on the part of a diligent claimant. On the face of the procedures, the process described by the District Court does not seem burdensome: one year would in the judgment of most be ample time to allow a claimant to respond to notice requesting a response. In addition, the VA is required to read any submission in the light most favorable to the claimant, and service representatives are available to see that various procedural steps are complied with. It may be that the service representative cannot, as the District Court hypothesized, provide all the services that a lawyer could, but there is no evidence in the record that they cannot or do not provide advice about time limits.
The District Court’s opinion is similarly short on definition or quantification of “complex” cases. If this term be understood to include all cases in which the claimant asserts injury from exposure to radiation or agent orange, only approximately 3 in 1,000 of the claims at the regional level and 2% of the appeals to the BVA involve such claims. Nor does it appear that all such claims would be complex by any fair definition of that term: at least 25% of all agent orange cases and 30% of the radiation cases, for example, are disposed of because the medical examination reveals no disability. What evidence does appear in the record indicates that the great *330majority of claims involve simple questions of fact, or medical questions relating to the degree of a claimant’s disability; the record also indicates that only the rare case turns on a question of law. There are undoubtedly “complex” cases pending before the VA, and they are undoubtedly a tiny fraction of the total cases pending. Neither the District Court’s opinion nor any matter in the record to which our attention has been directed tells us more than this.
The District Court’s treatment of the likely usefulness of attorneys is on the same plane with its efforts to quantify the likelihood of error under the present system. The court states several times in its opinion that lawyers could provide more services than claimants presently receive — a fact which may freely be conceded — but does not suggest how the availability of these services would reduce the likelihood of error in the run-of-the-mine case. Simple factual questions are capable of resolution in a nonadversarial context, and it is less than crystal clear why lawyers must be available to identify possible errors in medical judgment. Cf. Parham v. J. R., 442 U. S., at 609-612. The availability of particular lawyers’ services in so-called “complex” cases might be more of a factor in preventing error in such cases, but on this record we simply do not know how those cases should be defined or what percentage of all of the cases before the VA they make up. Even if the showing in the District Court had been much more favorable, appellees still would confront the constitutional hurdle posed by the principle enunciated in cases such as Mathews to the effect that a process must be judged by the generality of cases to which it applies, and therefore a process which is sufficient for the large majority of a group of claims is by constitutional definition sufficient for all of them. But here appellees have failed to make the very difficult factual showing necessary.12
*331Reliable evidence before the District Court showed that claimants represented by lawyers have a slightly better success rate before the BVA than do claimants represented by service representatives, and that both have a slightly better success rate than claimants who were not represented at all. Evidence also showed that there may be complex issues of causation in comparatively few of the hundreds of thousands of cases before the VA, but there is no adequate showing of the effect the availability of lawyers would have on the proper disposition of these cases. Neither the difference in success rate nor the existence of complexity in some cases is sufficient to warrant a conclusion that the right to retain and compensate an attorney in VA cases is a necessary element of procedural fairness under the Fifth Amendment.
*332We have in previous cases, of course, held not only that the Constitution permits retention of an attorney, but also that on occasion it requires the Government to provide the services of an attorney. The Sixth Amendment affords representation by counsel in all criminal proceedings, and in cases such as Gideon v. Wainwright, U. S. 335 (1963), and Argersinger v. Hamlin, 407 U. S. 25 (1972), we have held that this provision requires a State prosecuting an indigent to afford him legal representation for his defense. No one would gainsay that criminal proceedings are adversarial in nature, and of course the Sixth Amendment applies only to such proceedings.
In cases such as Gagnon v. Scarpelli, 411 U. S. 778 (1973), we observed that counsel can aid in identifying legal questions and presenting arguments, and that one charged with probation violation may have a right to counsel because of the liberty interest involved. We have also concluded after weighing the Mathews factors that the right to appointed counsel in a case involving the threatened termination of parental rights depends upon the circumstances of each particular case, see Lassiter v. Department of Social Services of Durham County, 452 U. S. 18 (1981), while three of the dissenters thought the same balancing required appointment of counsel in all such cases. Id., at 35 (Blackmun, J., joined by Brennan and Marshall, JJ., dissenting).
But where, as here, the only interest protected by the Due Process Clause is a property interest in the continued receipt of Government benefits, which interest is conferred and terminated in a nonadversary proceeding, these precedents are of only tangential relevance. Appellees rely on Goldberg v. Kelly, 397 U. S. 254 (1970), in which the Court held that a welfare recipient subject to possible termination of benefits was entitled to be represented by an attorney. The Court said that “counsel can help delineate the issues, present the factual contentions in an orderly manner, conduct cross-examination, and generally safeguard the interests of the *333recipient.” Id., at 270-271. But in defining the process required the Court also observed that “the crucial factor in this context... is that termination of aid pending resolution of a controversy over eligibility may deprive an eligible recipient of the very means by which to live while he waits. . . . His need to concentrate upon finding the means for daily subsistence, in turn, adversely affects his ability to seek redress from the welfare bureaucracy.” Id., at 264 (emphasis in original).
We think that the benefits at stake in VA proceedings, which are not granted on the basis of need, are more akin to the Social Security benefits involved in Mathews than they are to the welfare payments upon which the recipients in Goldberg depended for their daily subsistence. Just as this factor was dispositive in Mathews in the Court’s determination that no evidentiary hearing was required prior to a temporary deprivation of benefits, 424 U. S., at 342-343, so we think it is here determinative of the right to employ counsel. Indeed, there appears to have been no stated policy on the part of New York in Goldberg against permitting an applicant to divide up his welfare check with an attorney who had represented him in the proceeding; the procedures there simply prohibited personal appearance of the recipient with or without counsel and regardless of whether counsel was compensated, and in reaching its conclusion the Court relied on agency regulations allowing recipients to be represented by counsel under some circumstances. 424 U. S., at 342-343.
This case is further distinguishable from our prior decisions because the process here is not designed to operate adver-sarially. While counsel may well be needed to respond to opposing counsel or other forms of adversary in a trial-type proceeding, where as here no such adversary appears, and in addition a claimant or recipient is provided with substitute safeguards such as a competent representative, a decision-maker whose duty it is to aid the claimant, and significant concessions with respect to the claimant’s burden of proof, *334the need for counsel is considerably diminished. We have expressed similar concerns in other cases holding that counsel is not required in various proceedings that do not approximate trials, but instead are more informal and nonadversary. See Parham v. J. R., 442 U. S., at 608-609; Goss v. Lopez, 419 U. S. 565, 583 (1975); Wolff v. McDonnell, 418 U. S., at 570.
Thus none of our cases dealing with constitutionally required representation by counsel requires the conclusion reached by the District Court. Especially in light of the Government interests at stake, the evidence adduced before the District Court as to success rates in claims handled with or without lawyers shows no such great disparity as to warrant the inference that the congressional fee limitation under consideration here violates the Due Process Clause of the Fifth Amendment. What evidence we have been pointed to in the record regarding complex cases falls far short of the kind which would warrant upsetting Congress’ judgment that this is the manner in which it wishes claims for veterans’ benefits adjudicated. Schweiker v. McClure, 456 U. S. 188, 200 (1982); Mathews, 424 U. S., at 344, 349. The District Court abused its discretion in holding otherwise.
hH <
Finally, we must address appellees suggestion that the fee limitation violates their First Amendment rights. Appellees claim that cases such as Mine Workers v. Illinois State Bar Assn., 389 U. S. 217 (1967), and Railroad Trainmen v. Virginia ex rel. Virginia State Bar, 377 U. S. 1 (1964), establish for individuals and organizations a right to ensure “meaningful access to courts” for themselves or their members, and that the District Court was correct in holding that this right was violated by the fee limitation. There are numerous conceptual difficulties with extending the cited cases to cover the situation here; for example, those cases involved the rights of unions and union members to retain or recommend counsel *335for proceedings where counsel were allowed to appear, and the First Amendment interest at stake was primarily the right to associate collectively for the common good. In contrast, here the asserted First Amendment interest is primarily the individual interest in best prosecuting a claim, and the limitation challenged applies across-the-board to individuals and organizations alike.
But passing those problems, appellees’ First Amendment arguments, at base, are really inseparable from their due process claims. The thrust is that they have been denied “meaningful access to the courts” to present their claims. This must be based in some notion that VA claimants, who presently are allowed to speak in court, and to have someone speak for them, also have a First Amendment right to pay their surrogate speaker;13 beyond that questionable proposition, however, even as framed appellees’ argument recognizes that such a First Amendment interest would attach only in the absence of a “meaningful” alternative. The foregoing analysis of appellees’ due process claim focused on substantially the same question — whether the process allows a claimant to make a meaningful presentation — and we concluded that appellees had such an opportunity under the present claims process, and that significant Government interests favored the limitation on “speech” that appellees attack. Under those circumstances appellees’ First Amendment claim has no independent significance. The decision of the District Court is accordingly
Reversed.
A fourth individual veteran plaintiff died during the pendency of the proceedings.
Title 38 CFR §3.102 (1984) states:
“It is the defined and consistently applied policy of the Veterans Administration to administer the law under a broad interpretation, consistent, however, with the facts shown in every case. When, after careful consideration of all procurable and assembled data, a reasonable doubt arises regarding service origin, the degree of disability, or any other point, such doubt will be resolved in favor of the claimant. By reasonable doubt is meant one which exists by reason of the fact that the evidence does not satisfactorily prove or disprove the claim, yet a substantial doubt and one within the range of probability as distinguished from pure speculation or remote possibility. It is not a means of reconciling actual conflict or a contradiction in the evidence; the claimant is required to submit evidence sufficient to justify a belief in a fair and impartial mind that his claim is well grounded. Mere suspicion or doubt as to the truth of any statements submitted, as distinguished from impeachment or contradiction by evidence or known facts, is not a justifiable basis for denying the application of the reasonable doubt doctrine if the entire, complete record otherwise warrants involving this doctrine. The reasonable doubt doctrine is also applicable even in the absence of official records, particularly if the basic incident allegedly arose under combat, or similarly strenuous conditions, and is consistent with the probable results of such known hardships.”
Despite the general preclusion of judicial review with respect to VA benefits claims, this Court held in Johnson v. Robison, 415 U. S. 361 (1974), that the district courts have jurisdiction to entertain constitutional attacks on the operation of the claims systems.
The VA statistics show that 86% of all claimants are represented by service representatives, 12% proceed pro se, and 2% are represented by lawyers. App. 190. Counsel agreed at argument that the 12% who proceed pro se do so by their own choice.
The District Court rejected appellants’ argument that the question presented was controlled by this Court’s summary affirmance in Gendron v. Saxbe, 389 F Supp. 1303 (DC Cal.), summarily aff’d sub nom. Gendron v. Levi, 423 U. S. 802 (1975). Because we noted probable jurisdiction and heard oral argument in order to decide this case on the merits there is no need for us to determine whether the District Court properly distinguished Gendron.
The District Court in its opinion questioned “the extent to which it is possible to serve the interests of both the VA and claimants simultaneously,” and suggested that there was a “conflict” and that “the VA personnel might feel some pressure to protect the government purse. ” 589 F. Supp., at 1320, n. 17. There is no indication of such bias in the record-quite the contrary. Nor are we willing to accept that administrative adjudicators are presumptively subject to such bias.
Act of Aug. 24, 1937, ch. 754, §2, 50 Stat. 752, provided:
“In any suit or proceeding in any court of the United States to which the United States, or any agency thereof, or any officer or employee thereof, as such officer or employee, is a party, or in which the United States has intervened and become a party, and in which the decision is against the constitutionality of any Act of Congress, an appeal may be taken directly to the Supreme Court of the United States by the United States or any other party. ...”
The District Court held that applicants for benefits, no less than persons already receiving them, had a “legitimate claim of entitlement” to benefits if they met the statutory qualifications. The court noted that this Court has never so held, although this Court has held that a person receiving such benefits has a “property” interest in their continued receipt. See Atkins v. Parker, 472 U. S. 115, 128 (1985); Mathews v. Eldridge, 424 U. S. 319 (1976). Since at least one of the claimants here alleged a diminution of benefits already being received, however, we must in any event decide whether “due process” under the circumstances includes the right to be represented by employed counsel. In light of our decision on that question, infra, at 334, we need not presently define what class would be entitled to the process requested.
See Friendly, “Some Kind of Hearing,” 123 U. Pa. L. Rev. 1267, 1276 (1975):
“It should be realized that procedural requirements entail the expenditure of limited resources, that at some point the benefit to individuals from an additional safeguard is substantially outweighed by the cost of providing such protection, and that the expense of protecting those likely to be found undeserving will probably come out of the pockets of the deserving.”
Justice Stevens’ dissent quotes liberally from this same Senate Committee Report, -post, at 365-366, apparently intending to suggest that the Committee determined that the fee limitation was no longer justified. The quote is taken out of context, and as such it is quite misleading. The bill with respect to which the Report was issued would have provided for the first time for limited judicial review of BVA decisions. To this end, the Committee determined that “some easing of the limitation on attorneys’ fees” would be necessary to allow a claimant to pursue an effective appeal in the federal courts. But the proposed bill retained the fee limitation for all VA proceedings up to and including the first denial of a claim by the BVA. In the sections of the Report not quoted by Justice Stevens the Committee explained that the limitation was retained to “protect claimant’s benefits,” and because until judicial review was contemplated there was “no need” for attorneys. S. Rep. No. 97-466, p. 50 (1982). Finally, it is worth noting that in any event the proposed bill died in House Committee and thus was never enacted.
The District Court stated in its opinion that “both claimants and attorneys familiar with the VA system view that system as adversarial. ...” 589 F. Supp., at 1321. In reaching this conclusion, the District Court referred to statements by two attorneys and two claimants. One of the attorneys was admitted to practice in California in 1978, but does not take claims before the VA because of the fee limitation. His familiarity with VA procedures was acquired as a certified representative before the VA for appellee Swords to Ploughshares during his time as a law student. The second attorney was admitted to practice in Wisconsin in 1981, and has been a staff member of appellee Swords to Ploughshares since 1980. His representation of veterans has been primarily before discharge boards, but in the course of this representation he has become familiar with VA rules and practices. Both stated that they regarded the VA procedures as “adversarial.” Two claimants testified on the basis of their own experience, one that the VA had been “very adversarial” and the other that “the VA has opposed me at every turn. ...”
Anecdotal evidence such as this may well be sufficient to support a finding by a judge or jury in litigation between private parties that a particular fact did or did not exist. But when we deal with a massive benefits program provided by Congress in which 800,000 claims per year are decided by 58 regional offices, and 36,000 claims are appealed to the BVA, it is simply not the sort of evidence that will permit a conclusion that the entire system is operated contrary to its governing regulations.
Our understanding of the operation of the claims process is further bolstered by the findings of the Senate Committee alluded to earlier. As noted supra, at 322, that Committee conducted an extensive inquiry into *331the process in connection with several proposed bills that would have provided for judicial review of BVA decisions, and also would have withdrawn the fee limitation for proceedings occurring after the first denial by the BVA, while retaining the limitation for proceedings prior to that time. The Committee Report accompanying a 1982 bill noted its belief that the claims process presently operates informally and nonadversarially, that there was no evidence that most claimants were not satisfied "with the VA’s resolution of their claims, that there was in general “no need” for attorneys inasmuch as applying for benefits was a “relatively uncomplicated procedure,” and that the service organizations afforded a “high quality of representation.” S. Rep. No. 97-466, pp. 25, 49-50 (1982). Each bill unanimously passed the Senate, but died in House Committee, leaving the present system in operation. See S. 349, 97th Cong., 2d Sess (1982); S. 636, 98th Cong., 1st Sess. (1983).
When Congress makes findings on essentially factual issues such as these, those findings are of course entitled to a great deal of deference, inasmuch as Congress is an institution better equipped to amass and evaluate the vast amounts of data bearing on such an issue. See Rostker v. Goldberg, 453 U. S. 57, 72-73 (1981); Vance v. Bradley, 440 U. S. 93, 111-112 (1979); Katzenbach v. McClung, 379 U. S. 294 (1964). Because we do not believe the record in the District Court contradicted these findings, however, we need not rely on them, or determine what deference must be afforded on this congressional record; we mention the Committee’s findings only because they are entirely consistent with our understanding of the record developed in the District Court.
The dissent quotes from our decision in FEC v. National Conservative Political Action Committee, 470 U. S. 480, 493 (1985), post, at 364, n. 13, as if the analysis in that case answers the issues raised here. One would think that another proposition “so obvious that [it] seldom need[s] to be stated explicitly,” post, at 368, n. 16, is that the constitutional analysis of a regulation that restricts core political speech, such as the regulation at issue in FEC, will differ from the constitutional analysis of a restriction on the available resources of a claimant in Government benefit proceedings.