Justice Scalia
delivered the opinion of the Court.
In February 1958, James B. Stanley, a master sergeant in the Army stationed at Fort Knox, Kentucky, volunteered to participate in a program ostensibly designed to test the effectiveness of protective clothing and equipment as defenses against chemical warfare. He was released from his then-current duties and went to the Army’s Chemical Warfare Laboratories at the Aberdeen Proving Grounds in Maryland. Four times that month, Stanley was secretly administered doses of lysergic acid diethylamide (LSD), pursuant to an Army plan to study the effects of the drug on human subjects. According to his Second Amended Complaint (the allegations of which we accept for purposes of this decision), as a result of the LSD exposure, Stanley has suffered from hallucinations and periods of incoherence and memory loss, was impaired in his military performance, and would on occasion “awake from sleep at night and, without reason, violently beat his wife and children, later being unable to recall the entire incident.” App. 5. He was discharged from the Army in 1969. One year later, his marriage dissolved because of the personality changes wrought by the LSD.
On December 10, 1975, the Army sent Stanley a letter soliciting his cooperation in a study of the long-term effects of LSD on “volunteers who participated” in the 1958 tests. [672]*672This was the Government’s first notification to Stanley that he had been given LSD during his time in Maryland. After an administrative claim for compensation was denied by the Army, Stanley filed suit under the Federal Tort Claims Act (FTCA), 28 U. S. C. § 2671 et seq., alleging negligence in the administration, supervision, and subsequent monitoring of the drug testing program.
The District Court granted the Government’s motion for summary judgment, finding that Stanley “was at all times on active duty and participating in a bona fide Army program during the time the alleged negligence occurred,” No. 78-8141-Civ-CF, p. 2 (SD Fla., May 14, 1979), and that his FTCA suit was therefore barred by the doctrine of Feres v. United States, 340 U. S. 135 (1950), which determined that “the Government is not liable under the Federal Tort Claims Act for injuries to servicemen where the injuries arise out of or are in the course of activity incident to service.” Id., at 146. The Court of Appeals for the Fifth Circuit agreed that the Feres doctrine barred Stanley’s FTCA suit against the United States, but held that the District Court should have dismissed for lack of subject-matter jurisdiction rather than disposing of the case on the merits. Stanley v. CIA, 639 F. 2d 1146 (1981). The Government contended that a remand would be futile, because Feres would bar any claims that Stanley could raise either under the FTCA or directly under the Constitution against individual officers under Bivens v. Six Unknown Fed. Narcotics Agents, 403 U. S. 388 (1971). The court concluded, however, that Stanley “has at least a colorable constitutional claim based on Bivens,” 639 F. 2d, at 1159, and remanded “for the consideration of the trial court of any amendment which the appellant may offer, seeking to cure the jurisdictional defect.” Id., at 1159-1160.
Stanley then amended his complaint to add claims against unknown individual federal officers for violation of his constitutional rights. He also specifically alleged that the United States’ failure to warn, monitor, or treat him after he [673]*673was discharged constituted a separate tort which, because occurring subsequent to his discharge, was not “incident to service” within the Feres exception to the FTCA. See United States v. Brown, 348 U. S. 110 (1954). The District Court dismissed the FTCA claim because the alleged negligence was not “separate and distinct from any acts occurring before discharge, so as to give rise to a separate actionable tort not barred by the Feres doctrine.” 549 F. Supp. 327, 329 (SD Fla. 1982). It refused, however, to dismiss the Bivens claims. The court rejected, inter alia, the Government’s argument that the same considerations giving rise to the Feres exception to the FTCA should constitute “special factors” of the sort alluded to in Bivens, supra, at 396, and other cases as bars to a Bivens action. It cited as sole authority for that rejection the Court of Appeals for the Ninth Circuit’s decision in Wallace v. Chappell, 661 F. 2d 729 (1981). Sua sponte, the court certified its order for interlocutory appeal under 28 U. S. C. § 1292(b).
Following issuance of the order, the Government moved for partial final judgment pursuant to Federal Rule of Civil Procedure 54(b)1 on behalf of itself and three federal agencies that had (improperly) been named as FTCA defendants throughout the proceedings. The Government also argued that because no individual defendants had been named or served, and thus had neither appeared as parties nor sought representation from the Department of Justice, there was no one to seek interlocutory review of the court’s refusal to dismiss the Bivens actions. The court concluded that the Government’s contentions were “well taken,” Stanley v. CIA, 552 F. Supp. 619 (SD Fla. 1982), and on November 9, 1982, it granted the motion for partial final judgment, ordered the [674]*674Clerk to “enter final judgment in favor of the United States forthwith,” ibid., and vacated the portion of its prior order ruling on the Bivens claims against the individual defendants, giving Stanley 90 days to serve at least one individual defendant. The docket sheet for the case reflects the terms of that order (“The clerk to enter final judgment in favor of USA,” App. to Brief in Opposition A4), but does not indicate that an additional “separate document,” Fed. Rule Civ. Proc. 58, containing the judgment was entered. See Fed. Rule Civ. Proc. 79(a).
Stanley then filed his Second Amended Complaint, naming as defendants nine individuals (seven of whom are before us as petitioners) and the Board of Regents of the University of Maryland,2 and asserting civil rights claims under 42 U. S. C. §§ 1983 and 1985. Motions to dismiss for lack of-personal jurisdiction and improper venue were filed on behalf of some of the defendants (it was alleged that proper service had not been made on the others), but before those motions were ruled on, we issued our decision in Chappell v. Wallace, 462 U. S. 296 (1983), holding that “enlisted military personnel may not maintain a suit to recover damages from a superior officer for alleged constitutional violations,” id., at 305, and reversing the sole authority cited by the District Court in its prior order refusing to dismiss Stanley’s Bivens claims. Stanley’s counsel brought the Chappell decision to the attention of the District Court, which, apparently treating the fil[675]*675ing of the Second Amended Complaint as automatically reinstating its previously vacated order concerning the Bivens claims, sua sponte reconsidered and reaffirmed its prior decision. It concluded that, despite the broadly stated holding of the case, Chappell did not “totally ba[r] Bivens actions by servicemen for torts committed against them during their term of service.” 574 F. Supp. 474, 478 (1988). Rather, it said, Chappell only bars Bivens actions when “a member of the military brings a suit against a superior officer for wrongs which involve direct orders in the performance of military duty and the discipline and order necessary thereto,” 574 F. Supp., at 479, factors that in its view were not involved in Stanley’s claim. Nor could the court find in congressionally prescribed remedies, such as the Veterans’ Benefits Act, 38 U. S. C. §301 et seq., any expression of exclusivity of the sort Bivens contemplated would preclude recovery. See 403 U. S., at 397. The court again certified its order for interlocutory appeal under § 1292(b), which petitioners sought and the Court of Appeals for the Eleventh Circuit granted.
The Court of Appeals affirmed the conclusion that Chappell does not require dismissal of Stanley’s Bivens claims, on essentially the grounds relied upon by the District Court. 786 F. 2d 1490 (1986). The court did not think that Congress’ activity in the military justice field was a “special facto[r]” precluding Stanley’s claim, as “[t]hose intramilitary administrative procedures which the Court found adequate to redress the servicemen’s racial discrimination complaints in Chappell are clearly inadequate to compensate Stanley for the violations complained of here.” Id., at 1496.
Although the issue had not been addressed in the order from which the interlocutory appeal was taken, the Court of Appeals further determined that recent precedent in the Eleventh Circuit, including Johnson v. United States, 749 F. 2d 1530 (1985), rev’d, 481 U. S. 681 (1987), indicated that Stanley might have a viable FTC A claim against the United States, and that law-of-the-case principles therefore did not [676]*676require adherence to the 1982 holding that Stanley’s FTCA claim was barred by Feres. It remanded with instructions to the District Court to “allow Stanley the opportunity to amend to plead consistent with recent precedent.” 786 F. 2d, at 1499.
Because the Courts of Appeals have not been uniform in their interpretation of the holding in Chappell,3 and because the Court of Appeals’ reinstatement of Stanley’s FTCA claims seems at odds with sound judicial practice, we granted certiorari. 479 U. S. 1005 (1986).
I
We first address the Court of Appeals’ instruction to the District Court to allow Stanley to replead his FTCA claim. While petitioners advance several reasons why that action was improper, and additional reasons can perhaps be found in our recent decision in United States v. Johnson, 481 U. S. 681 (1987), we find it necessary to discuss only one. The case did not come before the Court of Appeals on appeal from a final decision of the District Court under 28 U. S. C. § 1291. Rather, the Court of Appeals had jurisdiction pursuant to § 1292(b), which provides:
“When a district judge in making in a civil action an order not otherwise appealable under this section, shall be of the opinion that such order involves a controlling question of law as to which there is substantial ground for difference of opinion and that an immediate appeal from the order may materially advance the ultimate termination of the litigation, he shall so state in writing in [677]*677such order. The Court of Appeals may thereupon, in its discretion, permit an appeal to be taken from such order . . . .” (Emphasis added.)
An appeal under this statute is from the certified order, not from any other orders that may have been entered in the case. Even if the Court of Appeals’ jurisdiction is not confined to the precise question certified by the lower court (because the statute brings the “order,” not the question, before the court), that jurisdiction is confined to the particular order appealed from. Commentators and courts have consistently observed that “the scope of the issues open to the court of appeals is closely limited to the order appealed from [and] [t]he court of appeals will not consider matters that were ruled upon in other orders.” 16 C. Wright, A. Miller, E. Cooper, & E. Gressman, Federal Practice and Procedure § 3929, p. 143 (1977). See Pritchard-Keang Nam Corp. v. Jaworski, 751 F. 2d 277, 281, n. 3 (CA8 1984), cert. dism’d, 472 U. S. 1022 (1985); United States v. Bear Marine Services, 696 F. 2d 1117, 1119, n. 1 (CA5 1983); Time, Inc. v. Ragano, 427 F. 2d 219, 221 (CA5 1970).
Here, the “order appealed from” was an order refusing to dismiss Stanley’s Bivens claims on the basis of our holding in Chappell. The Court of Appeals therefore had no jurisdiction to enter orders relating to Stanley’s long-dismissed FTCA claims, whether or not, as Stanley argues, “the issues involved in the Bivens claim and the alleged immunity of the individual defendants closely parallels [sic] the government’s immunity due to the Feres doctrine . . . [and] that is what all parties were arguing about in the interlocutory appeal.” Brief for Respondent 17-18. The Court of Appeals’ action is particularly astonishing in light of the fact that the United States was not even a party to the appeal, which involved only Stanley and the individual Bivens defendants (Stanley’s Bivens claim against the United States having been dis[678]*678missed by the District Court in 1982). We vacate that portion of the Court of Appeals’ judgment.4
II
That leaves the Court of Appeals’ ruling that Stanley can proceed with his Bivens claims notwithstanding the decision in Chappell. In our view, the court took an unduly narrow view of the circumstances in which courts should decline to permit nonstatutory damages actions for injuries arising out of military service.
In Bivens, we held that a search and seizure that violates .the Fourth Amendment can give rise to an action for damages against the offending federal officials even in the absence of a statute authorizing such relief. We suggested in dictum that inferring such an action directly from the Constitution might not be appropriate when there are “special factors counselling hesitation in the absence of affirmative action by Congress,” 403 U. S., at 396, or where there is an “explicit congressional declaration that persons injured by a federal officer’s violation of the Fourth Amendment may not recover money damages from the agents, but must instead be remitted to another remedy, equally effective in the view of Congress.” Id., at 397. We subsequently held that actions for damages could be brought directly under the Due Process Clause of the Fifth Amendment, Davis v. Passman, 442 U. S. 228 (1979), and under the Eighth Amendment’s proscription against cruel and unusual punishment, Carlson v. Green, 446 U. S. 14 (1980), repeating each time the dictum that “special factors counselling hesitation” or an “explicit congressional declaration” that another remedy is exclusive would bar such an action. 442 U. S., at 246-247; 446 U. S., at 18-19. In Chappell (and in Bush v. Lucas, 462 U. S. 367 [679]*679(1983), decided the same day), that dictum became holding. Chappell reversed a determination that no “special factors” barred a constitutional damages remedy on behalf of minority servicemen who alleged that because of their race their superior officers “failed to assign them desirable duties, threatened them, gave them low performance evaluations, and imposed penalties of unusual severity.” 462 U. S., at 297. We found “factors counselling hesitation” in “[t]he need for special regulations in relation to military discipline, and the consequent need and justification for a special and exclusive system of military justice . . . .” Id., at 300. We observed that the Constitution explicitly conferred upon Congress the power, inter alia, “[t]o make Rules for the Government and Regulation of the land and naval Forces,” U. S. Const. Art. I, § 8, cl. 14, thus showing that “the Constitution contemplated that the Legislative Branch have plenary control over rights, duties, and responsibilities in the framework of the Military Establishment. . . .” 462 U. S., at 301. Congress, we noted, had exercised that authority to “establish] a comprehensive internal system of justice to regulate military life, taking into account the special patterns that define the military structure.” Id., at 302. We concluded that “[t]aken together, the unique disciplinary structure of the Military Establishment and Congress’ activity in the field constitute ‘special factors’ which dictate that it would be inappropriate to provide enlisted military personnel a Bivenstype remedy against their superior officers.” Id., at 304.
Stanley seeks to distance himself from this holding in several ways. First, he argues that the defendants in this case were not Stanley’s superior military officers, and indeed may well have been civilian personnel, and that the chain-of-command concerns at the heart of Chappell and cases such as Gaspard v. United States, 713 F. 2d 1097, 1103-1104 (CA5 1983) (plaintiff was ordered to expose himself to radiation from nuclear test), cert. denied sub nom. Sheehan v. United States, 466 U. S. 975 (1984), are thus not implicated. Sec[680]*680ond, Stanley argues that there is no evidence that this injury was “incident to service,” because we do not know the precise character of the drug testing program, the titles and roles of the various individual defendants, or Stanley’s duty status when he was at the Maryland testing grounds. If that argument is sound, then even if Feres principles apply fully to Bivens actions, further proceedings are necessary to determine whether they apply to this case.
The second argument, however, is not available to Stanley here. The issue of service incidence, as that term is used in Feres, was decided adversely to him by the Court of Appeals in 1981, 639 F. 2d, at 1150-1153, and there is no warrant for reexamining that ruling here. See Allen v. McCurry, 449 U. S. 90, 94 (1980). As for his first argument, Stanley arid the lower courts may well be correct that Chappell implicated military chain-of-command concerns more directly than do the facts alleged here; in the posture of this case, one must assume that at least some of the defendants were not Stanley’s superior officers, and that he was not acting under orders from superior officers when he was administered LSD. It is therefore true that Chappell is not strictly controlling, in the sense that no holding can be broader than the facts before the court. It is even true that some of the language of Chappell, explicitly focusing on the officer-subordinate relationship that existed in the case at hand, would not be applicable here. To give controlling weight to those facts, however, is to ignore our plain statement in Chappell that “[t]he ‘special factors’ that bear on the propriety of respondents’ Bivens action also formed the basis of this Court’s decision in Feres v. United States,” 462 U. S., at 298, and that “[although this case concerns the limitations on the type of nonstatutory damages remedy recognized in Bivens, rather than Congress’ intent in enacting the Federal Tort Claims Act, the Court’s analysis in Feres guides our analysis in this case.” Id., at 299. Since Feres did not consider the officer-subordinate relationship crucial, but established instead an “incident to [681]*681service” test, it is plain that our reasoning in Chappell does not support the distinction Stanley would rely on.
As we implicitly recognized in Chappell, there are varying levels of generality at which one may apply “special factors” analysis. Most narrowly, one might require reason to believe that in the particular case the disciplinary structure of the military would be affected — thus not even excluding all officer-subordinate suits, but allowing, for example, suits for officer conduct so egregious that no responsible officer would feel exposed to suit in the performance of his duties. Somewhat more broadly, one might disallow Bivens actions whenever an officer-subordinate relationship underlies the suit. More broadly still, one might disallow them in the officer-subordinate situation and also beyond that situation when it affirmatively appears that military discipline would be affected. (This seems to be the position urged by Stanley.) Fourth, as we think appropriate, one might disallow Bivens actions whenever the injury arises out of activity “incident to service.” And finally, one might conceivably disallow them by servicemen entirely. Where one locates the rule along this spectrum depends upon how prophylactic one thinks -the prohibition should be (i. e., how much occasional, unintended impairment of military discipline one is willing to tolerate), which in turn depends upon how harmful and inappropriate judicial intrusion upon military discipline is thought to be. This is essentially a policy judgment, and there is no scientific or analytic demonstration of the right answer. Today, no more than when we wrote Chappell, do we see any reason why our judgment in the Bivens context should be any less protective of military concerns than it has been with respect to FTCA suits, where we adopted an “incident to service” rule. In fact, if anything we might have felt freer to compromise military concerns in the latter context, since we were confronted with an explicit congressional authorization for judicial involvement that was, on its face, unqualified; whereas here we are confronted with an explicit constitutional au[682]*682thorization for Congress “[t]o make Rules for the Government and Regulation of the land and naval Forces,” U. S. Const., Art. I, § 8, cl. 14, and rely upon inference for our own authority to allow money damages.5 This is not to say, as Justice Brennan’s dissent characterizes it, post, at 707, that all matters within congressional power are exempt from Bivens. What is distinctive here is the specificity of that technically superfluous grant of power,6 and the insistence (evident from the number of Clauses devoted to the subject) with which the Constitution confers authority over the Army, Navy, and militia upon the political branches. All this counsels hesitation in our creation of damages remedies in this field.
The othér major factor determining at which point, along the spectrum of generality, one should apply Chappell’s “special factors” analysis consists of the degree of disruption which each of them will in fact produce. This is an analytic rather than a policy judgment —but once again we see no reason why it should differ in the Bivens and the Feres contexts. Stanley underestimates the degree of disruption that would be caused by the rule he proposes. A test for liability that depends on the extent to which particular suits would call into question military discipline and decisionmaking would itself require judicial inquiry into, and hence intrusion upon, military matters. Whether a case implicates those concerns would often be problematic, raising the prospect of compelled depositions and trial testimony by military officers concern[683]*683ing the details of their military commands. Even putting aside the risk of erroneous judicial conclusions (which would becloud military decisionmaking), the mere process of arriving at correct conclusions would disrupt the military regime. The “incident to service” test, by contrast, provides a line that is relatively clear and that can be discerned with less extensive inquiry into military matters.
Contrary to the view of the Court of Appeals, 786 F. 2d, at 1496, it is irrelevant to a “special factors” analysis whether the laws currently on the books afford Stanley, or any other particular serviceman, an “adequate” federal remedy for his injuries. The “special facto[r]” that “counsels] hesitation” is not the fact that Congress has chosen to afford some manner of relief in the particular case, but the fact that congressionally uninvited intrusion into military affairs by the judiciary is inappropriate. Similarly irrelevant is the statement in Chappell, erroneously relied upon by Stanley and the lower courts, that we have “never held, nor do we now hold, that military personnel are barred from all redress in civilian courts for constitutional wrongs suffered in the course of military service.” 462 U. S., at 304. As the citations immediately following that statement suggest, it referred to redress designed to halt or prevent the constitutional violation rather than the award of money damages. See Brown v. Glines, 444 U. S. 348 (1980); Parker v. Levy, 417 U. S. 733 (1974); Frontiero v. Richardson, 411 U. S. 677 (1973). Such suits, like the case of Wilkes v. Dinsman, 7 How. 89 (1849), distinguished in Chappell, 462 U. S., at 305, n. 2, sought traditional forms of relief, and “did not ask the Court to imply a new kind of cause of action.” Ibid.
We therefore reaffirm the reasoning of Chappell that the “special factors counselling hesitation” — “the unique disciplinary structure of the Military Establishment and Congress’ activity in the field,” id., at 304 — extend beyond the situation in which an officer-subordinate relationship exists, and require abstention in the inferring of Bivens actions as exten[684]*684sive as the exception to the FTCA established by Feres and United States v. Johnson. We hold that no Bivens remedy is available for injuries that “arise out of or are in the course of activity incident to service.” 340 U. S., at 146.
Part II of Justice Brennan’s opinion argues in essence that because the refusal to entertain a Bivens action has the same effect as a grant of unqualified immunity, we should find “special factors” sufficient to preclude a Bivens action only when our immunity decisions would absolutely foreclose a money judgment against the defendant officials. The short answer to this argument is that Chappell made no reference to immunity principles, and Bivens itself explicitly distinguished the question of immunity from the question whether the Constitution directly provides the basis for a damages action against individual officers. 403 U. S., at 397. The analytic answer is that the availability of a damages action under the Constitution for particular injuries (those incurred in the course of military service) is a question logically distinct from immunity to such an action on the part of particular defendants. When liability is asserted under a statute, for example, no one would suggest that whether a cause of action exists should be determined by consulting the scope of common-law immunity enjoyed by actors in the area to which the statute pertains. Rather, one applies that immunity (unless the statute says otherwise) to whatever actions and remedies the terms of the statute are found to provide. Similarly, the Bivens inquiry in this case — whether a damages action for injury in the course of military service can be founded directly upon the Constitution — is analytically distinct from the question of official immunity from Bivens liability.
We do not understand Justice Brennan to dispute this. Rather, he argues that the answer to the former inquiry should be such that it produces a result coextensive with the answer to the latter. That is of course quite possible to achieve, since one can adjust the definition of a cause of ac[685]*685tion to produce precisely the same results as a given definition of immunity. For example, if a State wanted to eliminate driver liability for automobile accidents, it could either prescribe that all automobile drivers are immune from suit for injuries caused by their negligent driving or prescribe that no cause of action exists for injuries caused by negligent driving. But what Justice Brennan fails to produce is any reason for creating such an equivalency in the present case (and, presumably, in all Bivens actions). In the sole case he relies upon for his novel analysis, Davis v. Passman, 442 U. S. 228 (1979), there was a reason. There the Constitution itself contained an applicable immunity provision — the Speech or Debate Clause, Art. I, § 6, cl. 1—which rendered Members of Congress immune from suit for their legislative activity. The Court held that the “special concerns counseling hesitation” in the inference of Bivens actions in that area “are coextensive with the protections afforded by the Speech or Debate Clause.” 442 U. S., at 246. That is to say, the Framers addressed the special concerns in that field through an immunity provision — and had they believed further protection was necessary they would have expanded that immunity provision. It would therefore have distorted their plan to achieve the same effect as more expansive immunity by the device of denying a cause of action for injuries caused by Members of Congress where the constitutionally prescribed immunity does not apply.
Thus, Davis v. Passman would be relevant here if the Constitution contained a grant of immunity to military personnel similar to the Speech or Debate Clause. It does not, of course, and so we are compelled in the military field, as in others, to make our own assessment of whether, given the “special concerns counseling hesitation,” Bivens actions will lie. There is no more reason why court-created rules of immunity (as opposed to immunity specifically prescribed in the Constitution) should be held a priori to describe the limit of those concerns here than in any other field. Thus, the rule [686]*686Justice Brennan proposes is not an application but a repudiation of the “special factors” limitation upon the inference of Bivens actions. That limitation is quite hollow if it does nothing but duplicate pre-existing immunity from suit.
For the foregoing reasons, we vacate the Court of Appeals’ judgment that Stanley can assert an FTCA claim on remand to the District Court and reverse its judgment refusing to dismiss the Bivens claims against petitioners. The judgment of the Court of Appeals is reversed in part and vacated in part, and the case is remanded for further proceedings consistent with this opinion.
It is so ordered.
Justice Stevens joins Part I of this opinion.