dissenting:
In this case, the Bank and HRA, the developer, settled a financing dispute after a court action had been commenced. At the request of both parties, their settlement agreement was sealed by the court in February 1985.
In July 1985, some five months later, FAB III, a contractor who had furnished services to HRA, without having intervened in the Bank-HRA litigation, applied to the district court to unseal the Bank-HRA agreement. The district court denied FAB’s motion and refused to unseal the agreement holding that FAB’s claimed interest in access to the sealed information did not outweigh the public and private interests in favor of settling disputes.
The majority today, totally disregarding the difference between the sealing and the unsealing of an agreement, in an unprecedented decision, holds that the interest in settlement may not outweigh the public’s right of access even in private litigation. I cannot agree, and therefore, I dissent.
I.
I believe that the majority misconstrues the issue now before the court. The majority opinion claims the issue before us to be: “ ... under what circumstances [may] documents filed in the district court ... be sealed from public access.” Maj.Op., at 340. See also Maj.Op., at 345. However, that is not the issue this case presents. The question before us is not whether ma*347terial which is already public should now be sealed; the question on this appeal is whether a privately negotiated settlement agreement, agreed to and entered into a court record only on condition that it remain secret, should now be unsealed because of the district court’s supposed abuse of discretion in permitting it to be filed under seal.1
The resolution of the correct issue raised here dictates a different approach than the majority has taken: an approach that would protect the reliance interests of the parties absent exceptional circumstances. I therefore disagree with the majority’s decision to order the Bank-HRA settlement agreement terms unsealed.
Moreover, in reviewing the district court’s denial of FAB’s motion to unseal the Bank-HRA agreement, the majority adopts, for all practical purposes, a per se rule that the interest in settling cases can never outweigh the public’s right of access and thereby justify a court in sealing the terms of a voluntary settlement. The majority opinion holds “ the generalized interest in encouraging settlements does not rise to the level of interests that we have recognized may outweigh the public’s common law right of access. See Criden I, 648 F.2d at 829 ... [and] Smith I, 776 F.2d at 1113 ...” Maj.Op., at 346. The majority’s holding on this point is not compelled by any precedent (certainly not by Criden I or Smith I, which are the only authorities cited for that extraordinary proposition), and it utterly ignores the importance of, and the practical realities surrounding, the process of settling lawsuits.
Because I believe that the public and private interest in encouraging settlement is entitled to significant weight, and because the majority’s analysis and holding cannot help but impair seriously the efficacy of judicial efforts to encourage settlement of many cases, I cannot join in this result. For this reason, too, I respectfully dissent.
II.
As the majority recognizes, in deciding whether the common law right of access compels disclosure of materials before the court, the district court must “weigh[ ] the interests advanced by the parties in light of the public interest and the duty of the courts.” Nixon v. Warner Communications, 435 U.S. 589, 602, 98 S.Ct. 1306, 1314, 55 L.Ed.2d 570 (1978). This court has held that the common law right of access creates a presumption of access to all judicial records and documents. United States v. Smith, 776 F.2d 1104, 1110 (3d Cir.1985); United States v. Martin, 746 F.2d 964, 968 (3d Cir.1984). I therefore agree that, as a general matter, the common law right of access applies to settlement agreements when such agreements are filed with the court and become a part of the public record.
However, a settlement agreement that has never been disclosed to the public, but which was only entered into the record by the parties with the understanding that it would remain secret, presents a situation different from any situation that this court has addressed before. Although a presumption of access certainly arises when a court seals the transcript of a sidebar conference that has already taken place on the record in open court, United States v. Smith, 787 F.2d 111, 112 (3d Cir.1986), or when a party seeks access to material already entered into evidence and provided to the jury, United States v. Criden, 648 F.2d 814, 815 (3d Cir.1981), this case involves material and information that was never public, giving rise to a new and different factor: the reliance of the parties on the initial and continuing secrecy of the settlement agreement.
*348Although this court has not apparently addressed this precise situation, the Second Circuit has held that, “[o]nce a confidentiality order has been entered and relied upon, it can only be modified if an ‘extraordinary circumstance’ or ‘compelling need’ warrants the requested modification.” FDIC v. Ernst & Ernst, 677 F.2d 230, 232 (2d Cir.1982). The Second Circuit therefore affirmed the denial of a motion to lift a protective order sealing a settlement agreement.2 The Second Circuit recently reaffirmed its holding on this point. Palmi-eri v. State of New York, 779 F.2d 861, 864-65 (2d Cir.1985).
The Ernst & Ernst case, while not controlling in this court, presents what I regard as a sensible standard and a sensible result. Although the common law right of access must be given due regard, a court cannot operate in a vacuum. To apply mechanistically the same test no matter what the factual circumstances, is to risk doing injustice to parties before the court.
Therefore, I agree that when a document or item of evidence has been entered into the public record without any reliance on secrecy, the interests of the parties seeking to seal or unseal such material must be weighed in light of a presumption of openness. However, when a document has only been entered into the record in reliance on an order keeping it under seal, and when time has passed and the parties have acted in reliance on the terms of that settlement remaining under seal, I would hold with the Second Circuit that the presumption must shift. While the public interest in openness of court records must nevertheless be factored into the balance, I think it is appropriate that a protective order or seal order which itself induced the production or entry of the contested material be presumed to remain in effect, absent a showing of an “extraordinary circumstance” or “compelling need” by a third party seeking to unseal that information.
Indeed, in denying FAB Ill’s motion to unseal the settlement in this case, the district court cited Ernst & Ernst. See App. at 90. The majority opinion here also cites to Ernst & Ernst, but nowhere in its opinion does it address the Ernst & Ernst standard for unsealing.3
*349I would urge the adoption of the Ernst & Ernst standard and hold that where, as here, a settlement agreement has been filed with the court in reliance upon its being sealed, a third party who comes along after the fact bears the burden of establishing a “compelling need” or other “extraordinary circumstance” justifying access to the sealed agreement. I would further hold that because no such need or circumstance has been demonstrated, the burden has not been carried in this case, and the order of the district court should therefore be affirmed.
III.
Even if I were to accept (which I do not) the majority’s holding that the presumption in favor of access to judicial records still applies when a third party seeks to unseal a settlement agreement entered under seal and in reliance upon secrecy, I still could not agree to the majority’s rule of law that the interest in settling cases cannot serve to rebut the presumption of access and therefore cannot justify the sealing of a settlement agreement. See Maj.Op., at 345-347. I believe that such a rule is completely out of touch with the reality of running a trial court docket — a reality with which our district court judges must wrestle every day — and if permitted to remain as the law of our circuit will wreak havoc with judicial efforts to encourage settlement of appropriate cases.
Although I believe the matter to be self-evident, the majority’s out-of-hand rejection of the encouragement of settlement as a relevant factor in the decision to seal a settlement agreement requires me to explain both why I believe that fostering voluntary settlements of civil disputes is desirable and necessary, and why this goal will be unavoidably subverted by the majority’s holding in this case.
A.
It is impossible to gainsay that we have experienced a litigation explosion in the United States during the last twenty years, and that developing techniques for managing the increased caseloads and for otherwise stemming the burgeoning tide of litigation costs has become a subject of intense interest and debate. See generally, Symposium, Reducing the Costs of Civil Litigation, 37 Rutgers L.Rev. 217 (1985). Between 1973 and 1983, new filings of civil cases in the federal district courts rose from 98,560 to 241,842, an increase of 145 percent. Levin & Colliers, Containing the Cost of Litigation, 37 Rutgers L.Rev. 219, 27 n. 24 (1985). Perhaps more importantly, the number of long, complex, and difficult-to-try cases has also increased dramatically. The federal courts held 213 trials lasting 20 days or more in 1973. The figure doubled to 426 by 1983. Id. at 229.
To cope with the increasing volume of litigation, many commentators have advocated an active, “managerial” role for judges in supervising the course of litigation — a role that includes the encouragement of a variety of alternate means of resolving disputes short of full-dress trials. See Wall, Schiller & Ebert, Should Judges Grease the Slow Wheels of Justice? A Survey on the Effectiveness of Judicial Mediary Techniques, 8 Am.J. Trial Advoc. 83 (1984); Burger, Isn’t There a Better Way?, 68 A.B.A.J. 274 (1982).
Although arbitration, mediation, mini-trials, and other forms of alternative dispute resolution have gained prominence in recent years as potent weapons in the war against litigation glut, the key component of every rational approach to reducing the burden on our clogged court dockets has been and remains settlement. With very rare exceptions, see Fiss, Against Settle*350ment, 93 Yale L.J. 1073 (1984), commentators and judges who may concur on little else, agree on the value and necessity of a vigorous policy of encouraging fair and reasonable settlement of civil claims whenever possible. Indeed, the literature on the settlement of civil suits focuses not on whether settlement is desirable, but on how best to achieve it and how far a judge should go to encourage it. See, e.g., Provine, Settlement Strategies for Federal Judges (1986) (Federal Judicial Center Study); Craig & Christianson, The Settlement Process, 59 F.R.D. 203, 252 (1973); Fox, Settlement: Helping the Lawyers to Fulfill Their Responsibility, 53 F.R.D. 129 (1971).
This court, too, has recognized the overwhelming importance of settling civil suits and avoiding the wasted resources and institutional burden of trying every case:
Voluntary settlement of civil controversies is in high judicial favor. Judges and lawyers alike strive assidously to promote amicable adjustments of matters in dispute, as for the most wholesome of reasons they certainly should. When the effort is successful, the parties avoid the expense and delay incidental to litigation of the issues; the court is spared the burdens of a trial and the preparation and proceedings that must forerun it.
Pennwalt Corp. v. Plough, Inc., 676 F.2d 77, 80 (3d Cir.1982).
Indeed, recognition of the desirability of settlement has even found its way into the Federal Rules of Civil Procedure. Rule 16 was amended in 1983 to include the pursuit of settlement as an express goal of the pretrial conference. See Fed.R.Civ.P. 16(a)(5); 6 C. Wright & A. Miller, Federal Practice and Procedure § 1521 (West Supp. 1986). Thus, an activist role for judges in managing cases — and encouraging their settlement — has expressly been provided for under the federal rules.
As any trial judge knows, the settlement of civil cases is not just a permissible and desirable goal, but a practical necessity. In one study of cases filed in ten courts, fully 88 percent were settled; only nine percent went to trial. Galanter, Reading the Landscape of Disputes: What We Know and Don’t Know (and Think We Know) About Our Allegedly Contentious and Litigious Society, 31 U.C.L.A.L.Rev. 4, 28 (1983).
B.
While the importance of settlement would seem to be self-evident, I believe it is equally obvious that confidentiality is often a key ingredient in a settlement agreement — and that many settlements would not be reached if the secrecy of their terms could not be safeguarded.
Both courts and commentators have recognized the crucial role of confidentiality in facilitating settlement of civil cases. See Provine, Settlement Strategies, supra, at 34; Marcus, Myth and Reality in Protective Order Litigation, 69 Cornell L.Rev. 1, 28, 49 (1983). In addition, the need to keep secret the terms of a settlement has been recognized as a justification for imposing a protective order guaranteeing confidentiality of certain discovery material exchanged prior to settlement. Id. at 49, 52.
Indeed, the protective order affirmed by the Second Circuit in the Ernst & Ernst case (and discussed at length in the majority opinion at page 346) was expressly predicated upon the necessity of guaranteeing confidentiality as a condition of settlement. As the district court judge explained:
This was a complex, multi-district case involving numerous defendants and numerous plaintiffs. If an agreement could not be concluded, the trial would have lasted at least six months, necessitating enormous expenditures of both litigant and judicial resources. In addition to extensive costs already borne by the [parties] prior to the trial, litigation expenses likely would have consumed the balance of insurance protecting the individual defendants and providing some security for collection of the ultimate judgment.
All parties were cognizant of the fact that a critical factor in averting these *351untoward results was the element of con-fidentiality____ Without secrecy of the terms, a settlement would not have been consummated.
******
[T]he strong public policy favoring settlement of disputes, particularly in complicated cases, and the importance of the stability of judgments and settlements, argue strongly against modification of the order.
In re Franklin National Bank Securities Litigation, 92 F.R.D. 468, 472 (E.D.N.Y.1981), aff'd sub nom. FDIC v. Ernst & Ernst, 677 F.2d 230 (2d Cir.1982) (emphasis added).
While few cases address the question of the sealing of settlement agreements, I suspect that this is because many trial judges regard it as self-evident that secrecy is often necessary and they therefore order settlement agreements filed under seal as a matter of course. This conclusion is supported by the frequent references to such sealings made without comment or challenge in reported cases. See, e.g., Marine Midland Bank, N.A. v. Kilbane, 739 F.2d 958, 959 (4th Cir.1984); Owen v. United States, 713 F.2d 1461, 1462 (9th Cir. 1983); E.E.O.C. v. Strasburger, Price, Fel-ton, Martin and Unis, 626 F.2d 1272, 1274 (5th Cir.1980).
Parties may have many reasons for desiring secrecy for the terms of their settlements.4 Settlement agreements may include trade secrets or information that threaten the privacy of the parties. See Nixon v. Warner Communications, Inc., 435 U.S. 589, 598, 98 S.Ct. 1306, 1312, 55 L.Ed.2d 570 (1978). While this kind of information would itself justify a seal order, parties may in good faith be concerned about releasing a far wider range of information, including information which would not itself entitle the parties to a protective order, but which might stand in the way of settlement if required to be disclosed.
The necessity for confidentiality may be particularly acute in the mass tort area, where a defendant must look beyond the parameters of a settlement with a single plaintiff and anticipate the impact of its settlement on innumerable future cases. As Edward A. Dauer, Associate Dean of Yale Law School, explained in a Second Circuit Judicial Conference discussion on alternative dispute resolution:
There are legitimate, good faith reasons for the parties who are trying to work out a solution to something like this toxic tort case to want their discussions to be private, immune both from later admission and immune from discovery by other potential plaintiffs’ lawyers later down the road, maybe even from competitors, and I think there are good faith reasons for wanting that privacy. That confidentiality is a very large advantage that will, if it can be guaranteed, make these kind of [alternative] procedures even more useful as adjuncts to the judicial process than they already are.
101 F.R.D. 161, 233 (1983).
For example, if a defendant facing multiple plaintiffs seeks to settle a meritorious claim for a certain sum of money, it may be deterred from doing so if it knows that the terms of such a settlement would have to be made public. The defendant may reasonably assume that disclosure of the comparatively favorable settlement terms would interfere with its ability to settle other cases for smaller amounts. I have no doubt that if all such settlement details were by rule of law always public, many settlements would never take place at all. *352Many defendants would almost certainly proceed to trial rather than broadcast to all potential plaintiffs how much they might be willing to pay.
Moreover, it is precisely in the context of mass torts with multiple plaintiffs — such matters as air disasters, toxic injuries, and product liability claims — that the interest in settlement is particularly strong. Such cases are characteristically long, complex, and costly to try, and the savings in public and private resources achieved by settling them are immense. As one judge familiar with the trial of mass tort cases noted:
Even saving one week of judicial time per ease would, as most trial judges know, be substantial. For example, in the Daikon Shield litigation, the record disclosed that, if the usual percentage (90) of the 1000 members statewide class settled their cases, the savings of judicial resources in the trial of the remaining 100 would amount to 400 weeks, or, roughly eight years of trial time. In addition, there would be an estimated savings of $26 million in litigation expense to the parties and $7 million of court expenses.
Williams, Mass Tort Class Actions: Going, Going, Gone?, 98 F.R.D. 323, 328 (1983) (Footnotes omitted).
The rule announced today by the majority flies in the face of this reality.5 By holding that settlement agreements may not be sealed to serve the interest of encouraging settlements, I believe the majority has removed from the discretion of our district courts an important technique for encouraging settlement. It is my belief that district court judges in this circuit will read the majority’s opinion to the same effect.
Under the majority’s rule, a district court judge faced with the prospect of a six month, 12 month or longer trial, who is told by the parties that they would settle the case if the terms of settlement could be filed with the court under seal, would have only one choice — to reject the settlement and proceed to trial. The judge would have no discretion to accept the settlement under seal even if the proposed settlement contained little information of public interest and the interests of both sides of the dispute would be furthered by the settlement.
With all due respect to my colleagues in the majority, theirs is an illogical and impractical result. We are dealing here not with a constitutional right, but with a flexible common-law rule that has historically been applied subject to the discretion of the district court. Moreover, decisions regarding the management of its docket and the expediting of case resolutions would seem to lie at the core of the district court’s discretionary powers. Therefore, assuming that a district court judge correctly weighs the public and private interests involved in deciding whether to order a settlement admitted under seal, I can see no reason why (contrary to the majority rule) the interest in encouraging settlement should not be entitled to due weight.6 Indeed, any other rule would improperly abridge the traditional discretion of the district judge and seriously impair the ability of judges to expedite settlement.
IV.
Had the majority in this case recognized that the decision to seal and to unseal settlement agreements was to be left to the discretion of the district court employing *353the Nixon standard and burden of proof for sealing, and the Ernst & Ernst standard and burden of proof for unsealing —standards which the district court here utilized — I would have had far less reason to disagree with the majority result.
In this case, the district court acknowledged the correct standards and concluded in its order that “FAB Ill’s asserted interest in access to the sealed information does not outweigh the public and private interests in favor of settling disputes.” App. at SO. Although I believe that the district court thereby satisfied its duty to weigh the relevant factors, I could nevertheless understand how others might prefer a more detailed and particularized discussion of how the various asserted interests were, and were not, served on the facts before the court. In such a situation, however, the indicated resolution would have involved no more than a remand to the district court for a fuller statement of its reasons in denying FAB’s motion to unseal.
Unfortunately, the majority holds instead that the interest in settling cases cannot justify sealing a settlement agreement. Consistent with that holding the majority reverses the district court’s order.
I believe the adoption of such a rule can only be counter-productive and must necessarily have the effect of discouraging settlement in many cases that would otherwise be routinely ended by the parties’ agreement. This, in turn, will undoubtedly force costly and ultimately unnecessary trials.
Accordingly, I respectfully dissent from the majority’s judgment, and, rather than reverse, I would affirm the district court’s order which refused to unseal the Bank-HRA settlement agreement.
. Curiously, the majority opinion acknowledges the "unsealing" issue which the parties identified by stating, “We will construe the district court’s order as the parties have construed it ... i.e., as a refusal to unseal ... the settlement agreement." Maj.Op., at 342. However, having identified the true issue that was appealed, the majority opinion then fails to either discuss, address or resolve that issue in its opinion.
. In the district court opinion affirmed by the Second Circuit, now Chief Judge Weinstein wrote:
The settlement agreement resulted in the payment of substantial amounts of money and induced substantial changes in position by many parties in reliance on the condition of secrecy. For the court to induce such acts and then to decline to support the parties in their reliance would work an injustice on these litigants and make future settlements predicated upon confidentiality less likely.
In re Franklin National Bank Securities Litigation, 92 F.R.D. 468, 472 (E.D.N.Y.1981), aff'd sub nom. FDIC v. Ernst & Ernst, 677 F.2d 230 (2d Cir.1982).
. The majority has steadfastly declined to explain why it has refused to acknowledge the evident difference between the act of initial sealing under the Nixon standard — an act which involves no reliance — and the act of subsequent unsealing under the Ernst & Ernst standard, where the factor of the parties’ reliance is of primary concern. Instead the majority attempts to blur the significance of this distinction by stressing the timeliness of the unsealing motion brought by a third party and by claiming that "secret judicial proceedings” should not be countenanced as they will lead to the “overturn of centuries of tradition of open access” Maj. Op., at 345. I suggest that these are makeweight arguments.
First, it is not the timeliness of a third party’s "unsealing” application that matters. Rather, it is the reliance by the settling parties on the fact that the sealed materials will remain sealed that marks the essential difference between a court sealing a settlement agreement and a court, at some later date, granting an application to unseal that document. Because such settlements cannot readily or practicably be withdrawn. Ernst & Ernst requires that the parties seeking to unseal a relied upon sealed settlement, demonstrate extraordinary circumstances or a compelling need. Timeliness was not the thrust of Ernst & Ernst — reliance was.
The fact that in this case FAB waited five months — from February to July 1985 — before it applied to unseal the Bank-HRA agreement has little bearing on the fact that neither the Bank nor HRA would have entered into the settlement had they not relied upon the district court’s act in preventing disclosure of their agreement. Thus, the majority's attempt to distinguish Ernst & Ernst on the ground of "timeliness” is irrelevant.
Nor is it an answer to the Ernst & Ernst "unsealing” formula to characterize the sealing of documents as “secret judicial proceedings” *349tending to “overturn centuries of tradition.” Id. at 345. The district court employed the Nixon standard when it agreed to seal the Bank-HRA agreement. By balancing the various interests which Nixon requires a court to consider, the district court, in accordance with established Supreme Court precedent, determined that the Bank-HRA agreement should not be disclosed to third parties. I suggest that branding such an action as a "secret judicial proceeding," with all that such a term may connote, and claiming that sealing practices will "overturn centuries of tradition of open access” is not an adequate substitute for reasoned judicial analysis.
. The majority suggests that parties seeking to keep their settlement terms secret should simply stipulate on the record to dismiss the case while negotiating a private settlement. However, this approach is not always practicable, since the parties may, as in the present case, require the further involvement of the court in enforcing the terms of the settlement: a more efficient mechanism than bringing an entirely new suit. Moreover, the court may require terms of settlement to be submitted for its approval as to their fairness and reasonableness. In addition, many settlements, particularly in the mass tort and other contexts may not be implemented without the court’s assistance. This is especially so when payments and document exchanges are deferred over extended periods of time.
. The majority attempts to skirt this problem by purporting to leave open the issue of what, if any, showing would be necessary to override the strong presumption of access. Maj.Op., at 346. I do not read this attempt, and I seriously doubt whether district court judges will read this attempt, as diluting in any meaningful manner the majority's holding that a district court judge in this circuit henceforth will no longer be permitted to seal settlement agreements.
. Assuming that the district court made specific findings indicating that the interest in settlement was strongly served on the facts of this case, that the public interest in openness was not, and that the interests of the third party (in this case FAB III) in obtaining access to the settlement were not substantial, I believe it would be within the discretion of the district court to order that the settlement agreement remain under seal.