On this expedited interlocutory appeal, we review an order of the United States District Court for the Southern District of New York (George B. Daniels, Judge) de*166nying plaintiff Nickolas Zervos’s motion for an order preliminarily enjoining defendants Verizon New York, Inc. (“Verizon”), Empire HealthChoice (“Empire”), and United Healthcare Co., Inc. (“United”) from refusing to provide insurance coverage for a particular medical treatment.
We hold, inter alia, (1) that a district court’s decision to grant or deny a preliminary injunction is generally reviewed for abuse of discretion, and there is no exception to this rule for a case in which the district court has heard no live testimony; and (2) that the District Court, which heard no live testimony, did not abuse its discretion in denying Zervos’s motion for preliminary injunctive relief because Zer-vos had not shown — at least on the record as it currently stands — that there was either a likelihood that he would succeed on the merits of any of his claims, or that there were sufficiently serious questions as to the merits of his claims to be fair grounds for litigation.
Accordingly, we affirm the order of the District Court denying Zervos’s motion for a preliminary injunction.
I. BaCkground
Zervos is an employee of Verizon, and at all relevant times he has been entitled to health insurance under a plan (the “plan” or “Plan”) sponsored by Verizon, carried by Empire, and administered by United. In March 2000, Zervos was diagnosed with metastatic breast cancer, for which he underwent a radical mastectomy and was treated with conventional chemotherapy. This chemotherapy regimen concluded in October 2000, at which point Zervos’s treating physicians recommended that he undergo a different treatment — namely, a single cycle of high-dose chemotherapy, administered in conjunction with a transplant of bone marrow stem cells.1 (“HDCT”).
Zervos requested that Empire “pre-cer-tify” that it would pay for HDCT, but Empire refused to do so by letter dated *167October 18, 2000, stating that “[HDCT] is Experimental and Investigational in Breast Cancer Stage IV,” the stage to which Zervos’s breast cancer had apparently advanced.2 The October 18 letter invited Zervos to appeal, and he did so.
Empire then sent Zervos’s relevant records to an “external reviewer” — in this case, Dr. Thomas R. Spitzer, Director of the Bone Marrow Transplant Program and Deputy Chief of the Hematology-Oncology Unit at Massachusetts General Hospital. Dr. Spitzer filed a report that concluded that “there are no convincing data to show that [HDCT] is a superior approach to conventional chemotherapy alone for male patients with metastatic breast cancer.”
Empire abides by the decisions of its external reviewers regarding whether coverage should be provided in a particular case. Accordingly, by letter dated October 24, 2000, Empire denied Zervos’s appeal, explaining: “We are unable to authorize [HDCT] for the following reason: ‘There are no convincing data to show that [HDCT] is superior to conventional chemotherapy alone for males with metastatic breast cancer.’ [HDCT] remains denied as experimental and investigational.”
On January 29, 2001, Zervos initiated this action in the District Court. His complaint alleged violations of New York law, New York City law, and three federal statutes — the Employee Retirement Income Security Act of 1974 (“ERISA”), 29 U.S.C. §§ 1001 et seq.; Title VII of the Civil Rights Act of 1964 (“Title VII”), as amended, 42 U.S.C. §§ 2000e et seq.; and the Americans with Disabilities Act (“ADA”), 42 U.S.C. §§ 12112 et seq. After filing his complaint, Zervos moved by Order to Show Cause for an order “preliminarily enjoining Verizon, Empire, and/or United from denying and/or refusing to pre-certify and provide coverage [for HDCT to Zer-vos] ... on the grounds that it is ‘investi-gational’ or ‘experimental,’ or that ‘there are no convincing data to show that [HDCT] is superior to conventional chemotherapy alone for males with [metastatic breast cancer],’ or for any other reason it is otherwise excluded under his health and medical insurance plan, and compelling Empire to allow and promptly pay for [Zervos’s] claims for [HDCT] and associated procedures.”
Following a flurry of recusals, the cause was assigned to Judge Daniels. For the reasons set forth in a thorough Memorandum Opinion and Order, see Zervos v. Verizon N.Y., Inc., No. 01 Civ. 685, 2001 WL 253377 (S.D.N.Y. Mar.14, 2001), Judge Daniels denied Zervos’s motion for a preliminary injunction. This timely and expedited interlocutory appeal followed.
II. Discussion
A. Standard op Review
We have often stated — without qualification — that we review a district court’s decision on a motion for preliminary injunction for abuse of discretion. See, e.g., SG Cowen Sec. Corp. v. Messih, 224 F.3d 79, 81 (2d Cir.2000). Zervos argues, however, that our review is de novo “where the district court has consid*168ered strictly a documentary record, with no live testimony.” Appellant’s BRIEF at 3-4 (citing Donovan v. Bierwirth, 680 F.2d 263, 269-70 (2d Cir.1982)).
Before considering Zervos’s argument, we pause briefly to clarify the meaning of three important terms — de novo review, clear-error review, and abuse-of-discretion review.
De novo review is review without deference.3 See Salve Regina College v. Russell, 499 U.S. 225, 238, 111 S.Ct. 1217, 113 L.Ed.2d 190 (1991) (“When de novo review is compelled, no form of appellate deference is acceptable.”). When we review a district court’s decision de novo, we take note of it, and study the reasoning on which it is based. However, our review is independent and plenary; as the Latin term suggests, we look at the matter anew, as though it had come to the courts for the first time. See Black’s Law Diotionaey 435 (6th ed.1990) (defining “de novo” as “[a]new” and “afresh”).
“Clear error” is the standard under which appellate courts review a district court’s factual findings. See Ornelas v. United States, 517 U.S. 690, 694 n. 3, 116 S.Ct. 1657, 134 L.Ed.2d 911 (1996) (“ ‘Clear error’ is a term of art derived from Rule 52(a) of the Federal Rules of Civil Procedure, and applies when reviewing questions of fact.”). It is a deferential standard of review grounded, inter alia, on the belief that district courts have a good deal of “expertise” when it comes to fact-finding. Anderson v. City of Bessemer City, 470 U.S. 564, 574, 105 S.Ct. 1504, 84 L.Ed.2d 518 (1985). “A finding is ‘clearly erroneous’ when although there is evidence to support it, the reviewing court on the entire evidence is left with the definite and firm conviction that a mistake has been committed.” United States v. United States Gypsum Co., 333 U.S. 364, 395, 68 S.Ct. 525, 92 L.Ed. 746 (1948); accord City of Bessemer City, 470 U.S. at 573-74, 105 S.Ct. 1504 (“If the district court’s account of the evidence is plausible in light of the record viewed in its entirety, the court of appeals may not reverse it even though convinced that had it been sitting as the trier of fact, it would have weighed the evidence differently. Where there are two permissible views of the evidence, the fact-finder’s choice between them cannot be clearly erroneous.”).
Finally, there is abuse-of-discretion review.4 This is a second, more complicated, species of deferential appellate review. When a district court is vested *169with discretion as to a certain matter, it is not required by law to make a particular decision. Rather, the district court is empowered to make a decision — of its choosing — that falls within a range of permissible decisions. A district court “abuses” or “exceeds” the discretion accorded to it when (1) its decision .rests on an error of law (such as application of the wrong legal principle) or a clearly erroneous factual finding,5 or (2) its decision — though not necessarily the product of a legal error or a clearly erroneous factual finding — -cannot be located within the range of permissible decisions.6
With these definitions in mind, we turn to Zervos’s argument that there is an exception here to our customary abuse-of-discretion review. In Orvis v. Higgins, 180 F.2d 537 (2d Cir.1950) (Frank, J.), we held that the court of appeals may review without deference — that is de novo — not only a district court’s legal determinations, but its factual determinations as well, provided that such factual determinations are not based on live testimony (“the Orvis rule”). See id. at 539-40. A subsequent generation of cases extended the Orvis rule, holding that when a district court’s decision on a preliminary injunction motion is not based on live testimony, the court of appeals may review the decision itself— that is whether to grant or deny the preliminary injunction- — less deferentially than our customary abuse-of-discretion standard would otherwise permit (“the live *170testimony exception”). See State of N.Y. v. Nuclear Regulatory Comm’n, 550 F.2d 745, 751-52 (2d Cir.1977) (collecting cases that apply the live testimony exception and noting that this “line of cases ... follow[s] Orvis ”).
The rationale for the Orvis rule is that when a district judge’s factual determinations are based entirely on documentary evidence, “we can evaluate the written record as well as he can.” Citizens Comm. for Faraday Wood v. Lindsay, 507 F.2d 1065, 1066 n. 1 (2d Cir.1974). The basis for the live testimony exception applicable to review of preliminary injunction rulings is all but identical: “where th[e] [district] court did not hear live witnesses, whose credibility [generally plays] an essential part in its determination, and the case was decided on the basis of pleadings, affidavits and depositions, which the court of appeals is in as good a position as the district judge to read and interpret, the appellate court is not limited to reversal for abuse of discretion.” Donovan, 680 F.2d at 269.
It is thus apparent that the rationale for the Orvis rule suggested the live testimony exception. But the logic by which the live testimony exception was derived from the Orvis rule is a bit obscure, never (so far as we know) having been articulated. Indeed, it is not obvious how the Orvis rule (under which a district court’s factual determinations are reviewed de novo) might have compelled elaboration of the live testimony exception (under which a district court’s decision to grant or deny a preliminary injunction is reviewed de novo). The decision to grant or deny a preliminary injunction involves more than factual determinations (which under the Orvis rule are reviewed de novo) and legal conclusions (which are always reviewed de novo). It involves an exercise of discretion. See post at 174. This exercise of discretion is vested in the district court and reviewed for abuse, and nothing in the Orvis rule seems to suggest otherwise.
We need not attempt to illuminate this darkness. The question of how our Court extrapolated the live testimony exception from the Orvis rule is not currently relevant. Rather, what matters here is only this: As explained below, because the live testimony exception was based entirely on the Orvis rule, the demise of that rule means that the live testimony exception is no longer viable.
The Orvis rule was unambiguously repudiated by the 1985 amendments to Rule 52 of the Federal Rules of Civil Procedure (“the 1985 amendments”). See Fed. R.Crv.P. 52(a) (“Findings of fact, whether based on oral or documentary evidence, shall not be set aside unless clearly erroneous[.]”) (emphasis added); see also id., Advisory Committee Notes, 1985 Amendment (explaining that the 1985 amendments aimed to displace decisions — including our own — that held that no deference had to be accorded to district court factual determinations in cases in which live testimony was not heard because “the appellate court is in as good a position as the trial court to review a purely documentary record”). As the Supreme Court has noted, Orvis itself was one of the authorities that was swept aside in the wake of the 1985 amendments:
[Vjarious Courts of Appeals have on occasion asserted the theory that an appellate court may exercise de novo review over findings not based on credibility determinations. See, e.g., Orvis v. Higgins, 180 F.2d 537 (2d Cir.1950).... This theory has an impressive genealogy, having first been articulated in an opinion written by Judge Frank and subscribed to by Judge Augustus Hand, see Orvis v. Higgins, supra, but it is impossible to trace the theory’s lineage back to the text of Rule 52(a), which *171states straightforwardly that ‘findings of fact shall not be set aside unless clearly erroneous.’ That the Rule goes on to emphasize the special deference to be paid credibility determinations does not alter its clear command: Rule 52(a) does not make exceptions or purport to exclude certain categories of factual findings from the obligation of a court of appeals to accept a district court’s findings unless clearly erroneous.
City of Bessemer City, 470 U.S. at 574, 105 S.Ct. 1504 (selected citations and internal quotation marks omitted).
Indeed, in the years since the 1985 amendments became law, panels of this court have emphasized that the current Rule 52 is controlling—notwithstanding pre-1985 Circuit precedents that invoke the Orvis rule. See, e.g., Weissmann v. Freeman, 868 F.2d 1313, 1322 (2d Cir.1989). Moreover, we have not hesitated to disregard even posi-1985 Circuit precedents that apply the Orvis rule.7 See Paddington Corp. v. Attiki Importers & Distrib., Inc., 996 F.2d 577, 585 (2d Cir.1993); Bristol-Myers Squibb Co. v. McNeil-P.P.C., Inc., 973 F.2d 1033, 1044 (2d Cir.1992).
Because the now-displaced Orvis rule was the only prop that supported the live testimony exception, that exception has been irredeemably undermined. We therefore hold that a district court’s decision to grant or deny a preliminary injunction is not generally reviewed de novo; rather, such decisions are generally reviewed for abuse of discretion, and there is no exception to this rule for cases in which the district court heard no live testimony.
In reaching this conclusion, we are aware that decisions applying the live testimony exception have occupied something of a prominent place in our jurisprudence. See, e.g., Bridgeport Coalition for Fair Representation v. City of Bridgeport, 26 F.3d 271, 273-74 (2d Cir.) (declining to apply the live testimony exception because the district court held hearings) vacated on other grounds by 512 U.S. 1283, 115 S.Ct. 35, 129 L.Ed.2d 931 (1994). And we are of course mindful of the rule that one panel is ordinarily bound by prior panel decisions. See, e.g., Rocket Jewelry Box, Inc. v. Noble Gift Packaging, Inc., 157 F.3d 174, 176 (2d Cir.1998).
However, the principle of fidelity to prior panel decisions does not apply when the sole basis of those decisions—here, the Orvis rule—has been overruled by a change in the Federal Rules of Civil Procedure. See, e.g., Weissmann, 868 F.2d at 1322; cf. Reich v. D.M. Sabia Co., 90 F.3d 854, 858 (3d Cir.1996) (holding that a prior panel’s decision is not binding on a subsequent panel where a statute undermining the prior panel’s rationale was passed after the prior panel’s decision). See generally Williams v. Ashland Eng’g Co., 45 F.3d 588, 592 (1st Cir.1995) (“An existing panel decision may be undermined by controlling authority, subsequently announced .... ”), abrogated on other grounds by Carpenters Local Union No. *17226 v. United States Fid. & Guar. Co., 215 F.3d 136 (1st Cir.2000).
Moreover, we have often noted that one panel of this court is not bound by a prior panel decision whose “rationale is overruled, implicitly or expressly, by the Supreme Court.” United States v. Ianniello, 808 F.2d 184, 190 (2d Cir.1986) vacated on other grounds by United States v. Indelicato, 865 F.2d 1370, 1375 (2d Cir.1989). Such is the case here. The Supreme Court in Anderson made it clear that the Orvis rule has been toppled; law that has been built entirely atop it — such as the live testimony exception — must necessarily fall with it.
In sum, because the live testimony exception is no longer viable, we may displace the District Court’s decision to deny Zervos’s motion for a preliminary injunction only if we conclude that that decision amounted to an abuse of discretion.8 We now turn to that question.
B. Likelihood of Sucoess on the Merits/Sufficiently Serious Questions Going to the Merits
A party seeking a preliminary injunction ordinarily must show: (a) that it will suffer irreparable harm in the absence of an injunction and (b) either (i) a likelihood of success on the merits or (ii) sufficiently serious questions going to the merits to make them a fair ground for litigation and a balance of hardships tipping decidedly in the movant’s favor.9 See, e.g., Polymer Technology Corp. v. Mimran, 37 F.3d 74, 77-78 (2d Cir.1994).
We begin by considering the second set of factors — “likelihood of success on the merits” and “sufficiently serious questions going to the merits.” 10 As to these, our conclusions are straightforward.
*173First, for substantially the reasons stated by the District Court, we believe that Zervos has not yet demonstrated that there is even a remote possibility that he will be entitled to relief on the basis of his ADA claim. See Zervos, 2001 WL 253377, at *12.
Second, we are convinced that Zervos has not shown a likelihood of success on his Title VII claim — the crux of which is that Empire has discriminated on the basis of sex in its provision of coverage for HDCT. The District Court found that Empire has covered (or not covered) HDCT without regard to the sex of those that apply for it. See id. at *11. We perceive no clear error in this factual finding, and therefore we agree with the District Court that Zervos has not established that there is any chance that his Title VII claim may be meritorious. See id.
Finally, we agree with the District Court that Zervos has not demonstrated—at least on the administrative record, see Miller v. United Welfare Fund, 72 F.3d 1066, 1071 (2d Cir.1995) (“review [of a decision of a plan administrator] under the arbitrary and capricious standard is limited to the administrative record”) — that there is any realistic possibility that he may be entitled to relief on his ERISA claim. The District Court found that Empire’s Technological Assessment Committee (“TAC”) decided after a March 2000 meeting that it would no longer cover HDCT, and that TAC made this decision only after “reviewing] the scientific data available to them” and “considering all of the medical evidence of HDCT’s unproven benefit.” Id. at *9. Moreover, the TAC concluded, on an adequate record, that HDCT was not superior to conventional therapy, Empire denied coverage on this basis, and the District Court upheld the reasonableness of Empire’s decision on this ground.
In so ruling, the District Court did not abuse its discretion in concluding that Zervos’s ERISA claim is insufficiently substantial to support preliminary injunctive relief. The parties agree that Empire’s decision not to cover HDCT can be displaced only if it was arbitrary and capricious, and there is simply no basis in the record for concluding that such may have been the case here. Indeed, Empire’s decision to stop covering HDCT appears to have been both procedurally sound (in that it was based on the advice of TAC experts who consulted relevant scientific data) and substantively sound (in that “none of the evidence” supports the conclusion that HDCT is anything but “Experimental/ Investigational”). See generally Pagan v. NYNEX Pension Plan, 52 F.3d 438, 442 (2d Cir.1995) (“Under the arbitrary and capricious standard of review, we may overturn a decision to deny benefits only if it was without reason, unsupported by sub*174stantial evidence or erroneous as a matter of law”) (internal quotation marks omitted).
In sum, we hold that the District Court did not abuse its discretion in deciding that Zervos was not entitled to a preliminary injunction. That decision was not based on an error of law or on a clearly erroneous finding of fact. And because Zervos has not shown that there is a likelihood that he will succeed on the merits of any of his claims, or that there are “sufficiently serious questions going to the merits” of his claims so as to make them a “fair ground” for litigation, Polymer Tech. Corp., 37 F.3d at 77-78, we hold that the District Court’s decision not to enter a preliminary injunction in this case did not fall outside of the range of alternative decisions that it was permitted to make. See ante at 168-69. See generally Alliance Bond Fund, Inc. v. Grupo Mexicano de Desarrollo, S.A., 143 F.3d 688, 692-93 (2d Cir.1998) (“The [district] court is vested with full discretion to determine whether to grant ... [an] injunction and its scope.”) (citing Hecht Co. v. Bowles, 321 U.S. 321, 329, 64 S.Ct. 587, 88 L.Ed. 754 (1944)), reversed on other grounds by 527 U.S. 308, 119 S.Ct. 1961, 144 L.Ed.2d 319 (1999).
Before concluding, we pause here to commend Judge Daniels for ruling expeditiously on Zervos’s motion for a preliminary injunction. In this case — whatever it may ultimately require — justice can tolerate no substantial delays. Accordingly, we trust that if Zervos opts to proceed further, Judge Daniels will continue to treat this matter as an urgent one, and will impress upon the parties the need for prompt action.
Our decision today does not necessarily mean that Zervos may not be able to develop a successful federal claim. For example, Zervos suggested at oral argument that he hopes to determine in discovery whether Empire’s decision to deny his coverage request was tinged by a conflict of interest. If Empire was in fact operating under such a conflict, Zervos’s ERISA claim would necessarily stand on relatively more solid ground. See generally Firestone Tire. & Rubber Co. v. Bruch, 489 U.S. 101, 115, 109 S.Ct. 948, 103 L.Ed.2d 80 (1989) (“if a benefit plan gives discretion to an administrator or fiduciary who is operating under a conflict of interest, that conflict must be weighed as a factor in determining whether there is an abuse of discretion”) (internal quotation marks and brackets omitted). On such an issue, which is distinct from the reasonableness of the plan administrators’ decision, the district court will not be confined to the administrative record.
Similarly, Zervos may be able to establish that, under the plan, the focus of an “arbitrary and capricious” inquiry should not be on Empire’s initial decision to deny coverage for HDCT — that is, the decision that Empire made after the TAC meeting. Rather, Zervos may be able to show (1) that judicial review here should focus on the final, definitive decision to deny coverage (which was all but made by an external reviewer); (2) that in this case that decision was the one communicated to Zer-vos by letter dated October 24, 2000; and (3) that that decision was “arbitrary and capricious” because it was based on a material misunderstanding of what is meant under the plan by the term “Experimental/Investigational.” 11 Compare The Plan *175at 6 (defining “Experimental/ Investiga-tional” as “services ... which are not of proven benefit for the ... treatment of the Covered Person’s condition, or are not generally recognized by the medical community as effective or appropriate for that condition”) with Letter from Empire to Zervos (Oct. 24, 2000) (“We are unable to authorize [HDCT] for the following reason: ‘There are no convincing data to show that [HDCT] is superior to conventional chemotherapy alone for males with metastatic breast cancer.’ [HDCT] remains denied as experimental and investigational.’ ”) (emphasis added) and Letter from Thomas R. Spitzer to Mary E. Picerno (Oct. 20, 2000) (“[HDCT] appears to be as effective as prolonged combination chemotherapy for metastatic breast cancer”) (emphasis added). See generally Gallo v. Madera, 136 F.3d 326, 330 (2d Cir.1998) (“Even when trustees of a pension plan are entitled to deference in interpreting the terms of the plan, deference cannot be so broad as to permit them to graft additional requirements onto unambiguous plan definitions.”).
III. Conclusion
For the reasons stated above, the District Court’s order denying Zervos’s motion for a preliminary injunction is AFFIRMED. If another appeal is taken the Clerk of Court shall set an expedited briefing schedule, and the cause will then be heard promptly by this panel on letter briefs.
. HDCT has been described as follows:
[The treatment] is a procedure by which stem cells are harvested from the bone marrow of the patient’s body and purified of cancer cells. The patient is placed under general anesthesia while the bone marrow is extracted by needle. The bone marrow is then frozen and stored while the patient receives high, and potentially toxic, doses of chemotherapy. In some cases, the chemotherapy is administered in doses which exceed one thousand times the standard dosage for conventional chemotherapy treatment. This high dose chemotherapy kills not only the cancer, but also the patient's remaining bone marrow which produces white blood cells to protect the body from infection. The bone marrow, which is the most sensitive of all the body tissue, is also the most damaged by chemotherapy. After the chemotherapy is completed, the patient’s stored bone marrow is reinfused intravenously so that it may re-engraft. The bone marrow then multiplies rapidly to replace the marrow destroyed during the high-dose chemotherapy. Given that the bone marrow is the patient's own tissue, there is little danger of rejection. There are, however, significant dangers associated with administering high-dose chemotherapy without some additional treatment to regenerate the bone marrow. Because the toll on a patient's white blood cells is significant, the secondary treatment is essential to the patient's chances for survival. Thus, the bone marrow must be quickly reintroduced after high-dose chemotherapy treatment to "rescue” the patient from otherwise almost certain death.
Sharona Hoffman, A Proposal for Federal Legislation to Address Health Insurance Coverage for Experimental and Investigational Treatments, 78 Or L.Rev. 203, 211-12 (1999); see also, e.g., Edward A. Stadtmauer et al., Conventional Dose Chemotherapy Compared with High Dose Chemotherapy Plus Autologous Hematopoietic Stem Cell Transplantation for Metastatic Breast Cancer, 342 New Eng. J. Med. 1069, 1070 (2000) (describing HDCT in more technical terms).
. The referenced terms — “Experimental and Investigational” — are drawn from the plan, which provides in pertinent part that "[b]ene-fits are payable ... only where the ... treatment [sought] ... [is] required for the necessary treatment of ... Illness ... as distinct from those which are ... Experimental/Inves-tigational.” The plan defines “Experimental/Investigational” as "services ... which are not of proven benefit for the ... treatment of the Covered Person’s condition, or are not generally recognized by the medical community as effective or appropriate for that condition, as determined by the Claims Administrator.”
. Such review is ''traditionally” associated with appellate assessments of a district court’s legal conclusions. Pierce v. Underwood, 487 U.S. 552, 558, 108 S.Ct. 2541, 101 L.Ed.2d 490 (1988). However, other conclusions are reviewed de novo as well. See, e.g., FDIC v. Providence College, 115 F.3d 136, 140 (2d Cir.1997) ("mixed questions of law and fact are ... reviewed de novo ”); cf. post at 169-70 (noting that our Court used to review certain factual findings on an essentially de novo basis).
. In explicating this term, we are mindful of the need to tread lightly. "Abuse of discretion” is famously slippery — its meaning can vary between contexts, see Gasperini v. Center for Humanities, Inc., 149 F.3d 137, 141 (2d Cir.1998); see also Calderon v. Thompson, 523 U.S. 538, 567, 118 S.Ct. 1489, 140 L.Ed.2d 728 (1998) (Souter, J., dissenting) ("[T]he variety of subjects left to discretionary decision requires caution in synthesizing abuse of discretion cases.”), and there has been little consensus over the years as to precisely what the phrase means, see, e.g., Buffalo Courier-Express, Inc. v. Buffalo Evening News, Inc., 601 F.2d 48, 59 & n. 18 (2d Cir.1979) (Friendly, J.) (describing "radically different notions of the meaning of ... 'abuse of discretion’ ”). As Justice Frankfurter said about another much-used term, abuse of discretion is “a verbal coat of too many colors.” United States v. L.A. Tucker Truck Lines, Inc., 344 U.S. 33, 39, 73 S.Ct. 67, 97 L.Ed. 54 (1952) *169(Frankfurter, J., dissenting) (about "jurisdiction").
. See Cooter & Gell v. Hartmarx Corp., 496 U.S. 384, 405, 110 S.Ct. 2447, 110 L.Ed.2d 359 (1990) ("A district court would necessarily abuse its discretion if it based its ruling on an erroneous view of the law or on a clearly erroneous assessment of the evidence.”); Milanese v. Rust-Oleum Corp., 244 F.3d 104, 110 (2d Cir.2001) ("A district court abuses its discretion if it bases its ruling on a mistaken application of the law or a clearly erroneous finding of fact.").
. We have noted:
Discretion is said to be “abused” ("exceeded” would be both a more felicitous and correct term) when the decision reached is not within the range of decision-making authority a reviewing court determines is acceptable for a given set of facts. This determination that the range of acceptable decision-making has been exceeded in a particular case is assuredly one of law, but it is analytically distinct from a determination that a legal standard applicable to a generality of fact situations has been ignored, incorrectly applied, or inadequately applied in a particular case.
Stormy Clime Ltd. v. ProGroup, Inc., 809 F.2d 971, 974 (2d Cir.1987), overruled on other grounds by Two Pesos, Inc. v. Taco Cabana, Inc., 505 U.S. 763, 769, 112 S.Ct. 2753, 120 L.Ed.2d 615 (1992); see also Eastway Constr. Corp. v. City of N.Y., 821 F.2d 121, 123 (2d Cir.1987) ("All discretion is to be exercised within reasonable limits. The concept of discretion implies that a decision is lawful at any point within the outer limits of the range of choices appropriate to the issue at hand; at the same time, a decision outside those limits exceeds or, as it is infelicitously said, 'abuses' allowable discretion.”); United States Circuit Judge Joseph T. Sneed, Trial Court Discretion: Its Exercise by Trial Courts and its Review by Appellate Courts, Address to the Judges of the Second Circuit (Apr. 19, 1982) ("To have discretion is to have choice. To have choice is to be able to choose one course of action over one or more others with immunity from reversal by a higher court because of the course selected. The range of choice is determined by the number of permissible courses of action that exist.”). The great student of abuse of discretion review was the late Maurice Rosenberg, who at the time of his death was the Harold R. Medina Professor of Procedural Jurisprudence Emeritus at the Columbia University Law School. See Henry J. Friendly, Indiscretion About Discretion, 31 Emory L.J. 747, 756 n. 29 (1982) (explaining that he "borrowed outrageously” from Professor Rosenberg’s "splendid” work). For an excellent introduction to Professor Rosenberg’s scholarship, see Maurice Rosenberg, Appellate Review of Trial Court Discretion, 79 F.R.D. 173 (1975).
. Implicit references to the Orvis rule occasionally seep into our cases. In Brewer v. West Irondequoit Cent. Sch. Dist., 212 F.3d 738 (2d Cir.2000), for example, we stated that where “the district court makes the requisite findings based solely on a written record, we review the record de novo because we are in as good a position as the district court to read and interpret the submitted materials.” Id. at 744. In support of this proposition, Brewer relied on Innovative Health Systems, Inc. v. City of White Plains, 117 F.3d 37, 43 (2d Cir.1997), which in turn rested on Doe v. New York University, 666 F.2d 761, 765 (2d Cir.1981). Doe was based on Jack Kahn Music Co. v. Baldwin Piano & Organ Co., 604 F.2d 755, 758 (2d Cir.1979), and we have made it clear that Jack Kahn was superseded by the 1985 amendments. See Weissmann, 868 F.2d at 1322.
. There may be some circumstances in which we do not review a district court's preliminary injunction decision for abuse of discretion. See, e.g., Hsu v. Roslyn Union Free Sch. Dist. No. 3, 85 F.3d 839, 852-53 (2d Cir.1996) (suggesting that less deferential review may be appropriate when "the district court ruling resembles a final decision on the merits”). However, the only exception to our customary review for abuse of discretion invoked here by Zervos is the live testimony exception. And that exception, we have now held, is no longer extant.
. In certain circumstances, the movant must satisfy higher standards. See generally Tom Doherty Assocs., Inc. v. Saban Entertainment, Inc., 60 F.3d 27, 33-35 (2d Cir.1995) (describing such circumstances and standards). In this case, we hold that Zervos is not entitled to preliminary injunctive relief even under the relatively less demanding standards enumerated in the text. See post at 174. Accordingly, we need not and do not consider whether the higher standards described in Tom Doherty Associates have any bearing on this case.
. To be sure, in most cases "the moving party must first demonstrate that [irreparable] injury is likely before the other requirements for the issuance of an injunction will be considered.” Rodriguez v. DeBuono, 175 F.3d 227, 234-35 (2d Cir.1999) (emphasis added). Here, however, to determine whether non-issuance of a preliminary injunction would irreparably injure Zervos, we would need to ask and answer an arguably difficult question: Namely, to show irreparable injury in the context of a denial of coverage for medical treatment, must a preliminary injunction movant demonstrate that such treatment will help him, or does it suffice for the movant to show that there is a reasonable probability that the treatment will help him? Compare, e.g., Graham v. Medical Mut., 130 F.3d 293, 296 (7th Cir.1997) (holding that a preliminary injunction movant did not show irreparable injury because the evidence did not demonstrate that HDCT was more efficacious than the conventional chemotherapy that the mov-ant was receiving), with Marro v. K-III Comm. Corp., 943 F.Supp. 247, 249-50 (E.D.N.Y.1996) (holding that a preliminary injunction movant showed irreparable injury because the evidence demonstrated that if he did not receive HDCT his "only chance for long-term survival” would be lost, and stating that "[t]he issuance of an injunction compel*173ling [djefendant [insurance company] to pre-certify coverage, then, is literally a matter of life or death for [him]”) (emphasis added), and Henderson v. Bodine Aluminum, Inc., 70 F.3d 958, 961 (8th Cir.1995) ("[P]reliminary injunctions become easier to obtain as the plaintiff faces progressively graver harm. It is hard to imagine a greater harm than losing a chance for potentially life-saving medical treatment.”) (citation omitted, emphases added). To avoid unnecessarily reaching and resolving this question — and because we hold that Zervos was not entitled to a preliminary injunction even if its non-issuance would irreparably injure him, see post at 174 — we consider first whether he has shown either a likelihood of success on the merits of his claims or that there are "sufficiently serious” questions that go to the merits of those claims. See, e.g., Forest City Daly Housing, Inc. v. Town of North Hempstead, 175 F.3d 144, 153 (2d Cir.1999) (conducting first a "merits” analysis in a case in which (1) this analysis demonstrated that the movant was not entitled to preliminary injunctive relief and (2) an irreparable injury inquiry would have required resolution of difficult legal questions).
. We mention this argument not because we mean to intimate any views as to its merits, but rather because it is strongly suggested by Zervos’s brief, and because if it is pressed further by Zervos it will need to be addressed in the first instance — and relatively quickly— by the District Court.