Confirming and Vacating Arbitration Awards: The Narrow Door of FAA Review

Document type: Article Practice area: Litigation — Arbitration Jurisdiction: United States (federal) Last reviewed: 5 September 2026

There is a moment, familiar to anyone who has lost an arbitration, when the client asks the obvious question: can we appeal?

The honest answer is that you can try, and you will almost certainly lose. Empirical studies of federal vacatur petitions put the success rate in the range of five to ten percent, and the wins cluster in a narrow band of genuine procedural failures — an arbitrator who failed to disclose a business relationship with a party, a panel that decided a claim nobody submitted, an award that resolved nothing. "The arbitrator misread the contract" is not on the list. Neither is "the arbitrator ignored controlling precedent," at least not in most circuits, and not without more.

This is not a defect in the system. It is the system. Parties who agree to arbitrate are buying finality, and finality has a price: the possibility that the arbitrator will be wrong and no one will fix it. The Supreme Court has said as much, repeatedly, in language that leaves little room for creative advocacy. The Federal Arbitration Act gives courts a specific and short list of things they may do with an award, and courts that go beyond the list get reversed.

Understanding that list — precisely, not approximately — is the difference between advising a client sensibly and spending eighty thousand dollars on a motion that was never going to work.

The statutory architecture

The FAA's post-award provisions occupy four sections and about eight hundred words. They are worth reading in full at least once, because most of the confusion in this area comes from lawyers arguing about doctrines the statute does not contain.

Section 9 — confirmation. If the parties have agreed that a judgment may be entered on the award, any party may apply within one year for an order confirming it, and the court "must grant such an order unless the award is vacated, modified, or corrected as prescribed in sections 10 and 11." The word is must. Confirmation is not discretionary, and a court asked to confirm an unchallenged award has one job.

Section 10 — vacatur. The court may vacate on four grounds:

Where the award was procured by corruption, fraud, or undue means.
Where there was evident partiality or corruption in the arbitrators.
Where the arbitrators were guilty of misconduct — refusing to postpone a hearing on sufficient cause shown, refusing to hear evidence pertinent and material to the controversy, or any other misbehavior by which the rights of a party have been prejudiced.
Where the arbitrators exceeded their powers, or so imperfectly executed them that a mutual, final, and definite award upon the subject matter submitted was not made.

Section 11 — modification or correction. The court may modify or correct where there is an evident material miscalculation of figures or an evident material mistake in a description; where the arbitrators awarded on a matter not submitted to them, if it does not affect the merits of what was submitted; or where the award is imperfect in matter of form not affecting the merits.

Section 12 — the clock. Notice of a motion to vacate, modify, or correct must be served within three months after the award is filed or delivered.

That is the entire toolkit. Four vacatur grounds, three modification grounds, one deadline, and a mandatory confirmation provision. Everything else in this area is interpretation.

Hall Street: the grounds are exclusive

The most important structural decision is Hall Street Associates, L.L.C. v. Mattel, Inc., 552 U.S. 576 (2008), and its holding is short: the grounds in §§ 10 and 11 are exclusive, and parties cannot expand judicial review by contract.

The facts made the question unavoidable. Hall Street and Mattel were in a lease dispute over environmental indemnity. Mid-litigation, they agreed to arbitrate and drafted a review provision that any commercial lawyer would recognize as sensible risk management: the district court could vacate, modify, or correct the award if the arbitrator's findings of fact were not supported by substantial evidence, or if the conclusions of law were erroneous. The district court approved the agreement. The arbitrator ruled for Mattel. The district court, applying the contractual standard, found a legal error and sent it back. On the second pass the arbitrator ruled for Hall Street.

The Ninth Circuit held the expanded-review provision unenforceable, and the Supreme Court agreed. Section 10's text — "the United States court in and for the district wherein the award was made may make an order vacating the award upon the application of any party to the arbitration ... [in the enumerated circumstances]" — describes a closed set. The statutory scheme, the Court reasoned, "substantiat[es] a national policy favoring arbitration with just the limited review needed to maintain arbitration's essential virtue of resolving disputes straightaway." Parties who want appellate review can litigate; they cannot have arbitration's speed and litigation's error correction at the same time.

Two consequences follow, and both matter in practice.

First, expanded-review clauses do not work under the FAA. They still appear in agreements, often drafted by lawyers who have not read Hall Street, and they are unenforceable in the federal courts. Some states permit expanded review under state arbitration statutes, which creates a genuine drafting choice — but it is a choice about which arbitration law governs, not a clause you can add to an FAA arbitration.

Second, the status of "manifest disregard of the law" became genuinely unsettled. Before Hall Street, most circuits recognized manifest disregard as a judicially created vacatur ground, drawn from a passing reference in Wilko v. Swan. Hall Street said the FAA grounds are exclusive and described the Wilko language as possibly "shorthand" for the statutory grounds. The circuits split. Some treat manifest disregard as abrogated. Some treat it as surviving as a judicial gloss on § 10(a)(4). Some have not squarely decided. Nowhere is it a route to relief in a case where the arbitrator simply got the law wrong: even in circuits that retain the doctrine, it requires that the arbitrator knew of a governing legal principle, that the principle was well defined and explicitly applicable, and that the arbitrator deliberately refused to apply it. That is a mental-state showing, and arbitral awards almost never contain the evidence for it.

"Exceeded their powers" is not a merits review

Section 10(a)(4) is where most vacatur petitions live, because it sounds broad. It is not, and the Supreme Court has said so in terms that leave little wiggle room.

The controlling formulation comes from Oxford Health Plans LLC v. Sutter, 569 U.S. 564 (2013). An arbitrator had construed a clause with no reference to class procedures as authorizing class arbitration. Oxford argued the arbitrator exceeded his powers. The Court affirmed the award, and Justice Kagan's opinion contains the sentence every arbitration lawyer should have memorized:

"The sole question for us is whether the arbitrator (even arguably) interpreted the parties' contract, not whether he got its meaning right or wrong."

And, more bluntly:

"The potential for those mistakes is the price of agreeing to arbitration. ... The arbitrator's construction holds, however good, bad, or ugly."

That is the standard. Did the arbitrator interpret the contract? If yes, the inquiry ends, regardless of how implausible the interpretation is. A party seeking vacatur under § 10(a)(4) "bears a heavy burden," and it is not enough "to show that the panel committed an error — or even a serious error."

Contrast Stolt-Nielsen S.A. v. AnimalFeeds International Corp., 559 U.S. 662 (2010), where the Court did vacate. The parties had stipulated that their agreement was silent on class arbitration. The panel nonetheless imposed class arbitration, reasoning from public policy rather than from the agreement. The Court held the panel exceeded its powers because it did not interpret the contract at all — it "simply imposed its own view of sound policy."

The line between Stolt-Nielsen and Oxford Health is the line that decides most § 10(a)(4) motions. An arbitrator who interprets badly is safe. An arbitrator who does not interpret at all is not. In practice, the second category is rare, because reasoned awards almost always quote the contract somewhere, and a citation to the contract is usually enough to establish that interpretation occurred.

Major League Baseball Players Association v. Garvey, 532 U.S. 504 (2001) makes the same point from the other direction. The Ninth Circuit had vacated an arbitrator's factual findings and then, remarkably, directed the entry of an award for Garvey. The Supreme Court reversed summarily, reiterating that "courts are not authorized to review the arbitrator's decision on the merits despite allegations that the decision rests on factual errors or misinterprets the parties' agreement," and that "even serious error" does not suffice. The labor-arbitration cases — United Paperworkers International Union v. Misco, Inc., 484 U.S. 29 (1987) and United Steelworkers v. Enterprise Wheel & Car Corp., 363 U.S. 593 (1960) — supply the same rule in the collective bargaining context: so long as the award "draws its essence from the collective bargaining agreement," it stands.

Evident partiality: the ground that actually works

If there is a live vacatur ground, it is § 10(a)(2). Disclosure failures are the most common route to a vacated award, and unlike the merits grounds, they turn on facts that can be developed.

The foundational case is Commonwealth Coatings Corp. v. Continental Casualty Co., 393 U.S. 145 (1968). The neutral arbitrator in a three-member panel had an ongoing business relationship with one party — sporadic engineering consulting work, including on the very projects at issue, generating roughly $12,000 in fees. He disclosed nothing. The Court vacated.

The opinion is famously fractured. Justice Black's plurality suggested that arbitrators must disclose "any dealings that might create an impression of possible bias," a standard as broad as it sounds. Justice White, concurring in a two-Justice opinion that lower courts have treated as controlling, wrote more narrowly: arbitrators are not automatically disqualified by a business relationship if the parties are informed, and "arbitrators are not to be held to the standards of judicial decorum of Article III judges." The circuits have accordingly developed varying formulations — some requiring a showing that a reasonable person would conclude the arbitrator was partial, others focusing on whether the undisclosed relationship was material and substantial.

What survives across the formulations:

Nondisclosure of a material relationship is the classic case. An arbitrator with an undisclosed ongoing financial relationship with a party or its counsel is the paradigm.
The duty is continuing. Relationships that arise during the arbitration must be disclosed when they arise, not at the end.
Trivial or attenuated contacts do not suffice. Membership in the same bar association, an appearance at the same conference, a lawyer who once opposed the arbitrator in an unrelated case fifteen years ago — none of these get you there.
Waiver is real. A party that learns of a relationship during the proceeding and says nothing has almost certainly waived the objection. This is the single most common way an otherwise viable partiality challenge dies.

The practical implication for counsel: run conflicts on the arbitrator, thoroughly, at the outset, and again if the panel composition changes. Search the arbitrator's firm, prior awards, public filings, and the client's own vendor and litigation records. If something turns up mid-arbitration, raise it immediately and in writing. A partiality objection preserved on the record at the time is a real motion; the same objection raised for the first time after an adverse award is a waiver argument you will lose.

Section 10(a)(3): refusing evidence and refusing postponement

The misconduct ground is narrower than it reads. Arbitrators have broad discretion over procedure — that is much of what parties buy — and courts are reluctant to second-guess evidentiary rulings.

"Refusing to hear evidence pertinent and material" requires more than an adverse evidentiary ruling. The successful cases involve a panel that excluded an entire category of evidence going to a dispositive issue, or refused to permit any testimony from a key witness, in circumstances where the exclusion deprived a party of a fair hearing. A panel that limited a party to four depositions instead of eight, or declined to admit a late-produced document, is exercising discretion.

"Refusing to postpone the hearing upon sufficient cause shown" works in a similar band. Denying a continuance where a party's principal witness was hospitalized may be misconduct; denying a fourth continuance requested three days before a hearing set eight months earlier is not.

"Other misbehavior by which the rights of any party have been prejudiced" is a catch-all courts read narrowly. Ex parte communications with a party on the merits, a panel that conducted an unannounced site visit and relied on it, a panel that decided the case on a theory neither side briefed without giving notice — these are the shapes that occasionally succeed.

The common thread is fundamental fairness, not correctness. The question courts ask is whether the party got a hearing, not whether it got the hearing it wanted.

Badgerow and the jurisdictional trap

The most consequential recent development is not about the merits at all. It is about which court can hear a confirmation or vacatur petition, and it caught a great many practitioners by surprise.

The FAA is famously "something of an anomaly" in the federal statutory scheme: it creates substantive rights but confers no independent federal subject matter jurisdiction. A party seeking relief under the FAA must find jurisdiction somewhere else.

For petitions to compel arbitration under § 4, Vaden v. Discover Bank, 556 U.S. 49 (2009) held that a court may "look through" the petition to the underlying substantive dispute. If the underlying controversy arises under federal law, the § 4 petition may be heard in federal court. The textual hook was § 4's instruction that a court may hear the petition if it "would have jurisdiction ... of a suit arising out of the controversy between the parties."

Badgerow v. Walters, 596 U.S. 1 (2022) held that the look-through approach does not extend to petitions to confirm or vacate under §§ 9 and 10 — because those sections contain no comparable language. The Court's reasoning was straightforwardly textual: Congress included the "save for" clause in § 4 and omitted it from §§ 9, 10, and 11, and that difference must be given effect.

The practical consequence is large. A dispute over federal statutory claims — securities, antitrust, employment discrimination, patent license royalties — may be compelled to arbitration in federal court under § 4. When the award issues, the confirmation petition often cannot be filed in that same federal court, because the petition itself presents only a state-law contract question (enforce this award) between parties who may not be diverse.

What practitioners do about it:

Confirm in state court. Most states have confirmation procedures, and state courts routinely confirm FAA awards. This is now the default in many matters.
Use diversity. Where the parties are diverse and the award exceeds $75,000, § 1332 supplies jurisdiction independent of the subject matter. Note that the amount in controversy in a confirmation petition is generally the amount of the award.
Keep the compelled case open. Where a federal court compelled arbitration and stayed rather than dismissed the action, it retains the case, and a motion to confirm in that action stands on different footing than a freestanding petition. This is one reason Smith v. Spizzirri, 601 U.S. 472 (2024) — holding that § 3 requires a stay rather than dismissal when a party requests one — matters more than it first appears. Ask for the stay.
Draft for it. Some agreements now specify that any petition to confirm or vacate shall be brought in a designated state court, which at least removes the surprise.

The trap to avoid: filing a federal confirmation petition on a look-through theory, having it dismissed for lack of jurisdiction, and discovering that the three-month vacatur window in § 12 has run in the meantime — or that the one-year confirmation window in § 9 has. File in a court that has jurisdiction the first time.

The three-month clock

Section 12 requires that notice of a motion to vacate, modify, or correct be served within three months after the award is filed or delivered. Courts treat this as a hard deadline.

Three points that recur:

It is not tolled by settlement discussions, by requests to the arbitrator for clarification, or by the pendency of a related proceeding. Parties negotiating after an award routinely let the window close and then discover they have no leverage.

Most circuits hold that a party who misses the deadline cannot raise § 10 grounds defensively in opposition to a confirmation petition. The rule is not universal, and there is some authority the other way, but the safe assumption is that the three months are your only chance.

The clock runs from delivery of the award, not from a later corrected or clarified award — usually. Where a party timely seeks correction under the applicable arbitral rules and the panel issues a modified award, some courts run the clock from the modified award. Do not rely on it. Calendar three months from the original award and treat any extension as a bonus.

Practice point: the day the award arrives, calendar three separate dates — the three-month vacatur deadline, the one-year confirmation deadline under § 9, and, if the award is a foreign or nondomestic award under Chapter 2, the three-year confirmation deadline under § 207. Then decide, within the first two weeks, whether a challenge is worth evaluating at all.

What happens after vacatur

Vacatur is not victory; it is a second arbitration. Under § 10(b), where the time for the agreement has not expired, the court may direct a rehearing by the arbitrators.

Three practical questions arise:

Does the same panel rehear it? Often yes, unless the vacatur ground was partiality or misconduct by a specific arbitrator, in which case the arbitrator is replaced under § 5.

What is the scope of the rehearing? A court vacating on a narrow ground may limit the remand — for example, to redetermining damages after excluding an improperly considered theory. Ask for the limitation expressly; a general vacatur reopens everything.

Is the doctrine of functus officio a problem? Once an arbitrator issues a final award, the arbitrator's authority is generally exhausted and the arbitrator cannot revisit it. Exceptions exist for correcting clerical errors, for completing an award that omitted a submitted issue, and for clarifying genuine ambiguity. But a party that wants the arbitrator to reconsider the merits after the award has, in most cases, no vehicle. This is why the post-award correction procedures in the institutional rules — and their short internal deadlines, often twenty to thirty days — matter so much.

The reasoned award problem

Arbitral awards come in several forms, and the form the parties chose has real consequences for both confirmation and challenge.

A standard or bare award states the outcome — respondent shall pay claimant $2.4 million — without explanation. It is close to unreviewable, which is exactly what some parties want. It is also close to unusable if you need to know what was decided for preclusion purposes, for an indemnity claim against a third party, or for a related proceeding.

A reasoned award explains the basis. It is what most commercial parties elect, and it is the default in many institutional rules for larger cases. It provides the record needed to establish that the arbitrator interpreted the contract — which, after Oxford Health, is usually fatal to a § 10(a)(4) challenge. It is therefore, somewhat counterintuitively, protective of the award: the very reasoning a losing party hopes to attack is the reasoning that demonstrates the arbitrator did the job.

Findings of fact and conclusions of law are the most detailed form, sometimes elected in high-value matters. They give the most material to attack and the most evidence that no attack will succeed.

The drafting choice: if the parties value finality above all, a standard award serves. If they anticipate needing the award for a downstream purpose — collateral estoppel, insurance, contribution, tax — a reasoned award is worth the modest additional cost and time. Specify the form in the arbitration clause or, at the latest, in the initial procedural order. Arguing about it after the hearing is too late.

A worked case: the Halloran Ridge award

Halloran Ridge Energy and Castellane Turbines arbitrated a dispute over eleven wind turbines that underperformed their warranted capacity factor. The contract had an AAA commercial clause, a three-arbitrator panel, a reasoned award, and a limitation of liability capping damages at the contract price.

The panel awarded Halloran Ridge $18.4 million — roughly $3 million above the contract price — reasoning that the limitation clause did not apply because Castellane's conduct constituted a "fundamental breach" that, in the panel's view, rendered the limitation unenforceable under governing law.

Castellane's litigation counsel, Nkechi Vandersloot-Amari, worked the problem in this order.

Step one: the clock

The award was delivered on 14 April. She calendared 14 July as the § 12 deadline and worked backward: a decision by 30 May, a drafted motion by 20 June, service by 7 July. She also calendared the AAA's twenty-day window for requesting correction of computational errors, which expired 4 May — and decided not to use it, because the panel's error was substantive, not computational, and a correction request would have accomplished nothing while signaling the challenge.

Step two: the ground

Section 10(a)(4) — exceeded powers. The argument: the panel awarded damages above a contractual cap, and a panel that disregards an express limitation of liability is not interpreting the contract, it is rewriting it.

Nkechi read the award again with Oxford Health in front of her. The panel had quoted the limitation clause, described the parties' competing readings of it, cited the governing law on when limitations are unenforceable, and explained why it concluded the clause did not bar the award. That is interpretation. It may be wrong interpretation — she thought it was — but under Oxford Health the sole question is whether the panel "even arguably" interpreted the contract, and it plainly had.

This was not Stolt-Nielsen, where the panel had imposed a result from policy while the parties had stipulated the contract was silent. The panel here reasoned from the contract to a conclusion Castellane disliked.

Her assessment: 10 to 15 percent.

Step three: the other grounds

Evident partiality. She ran fresh conflicts. One panel member's firm had represented an affiliate of Halloran Ridge's parent in an unrelated matter that concluded four years before the arbitration. It had been disclosed in the arbitrator's initial disclosure statement. Castellane had not objected. Under Commonwealth Coatings and every circuit's gloss on it, a disclosed, concluded, unrelated representation is not evident partiality, and the failure to object at the time waived what little there was. Not viable.

Section 10(a)(3). The panel had denied Castellane's request for a second expert on turbine aerodynamics and limited each side to six depositions. Both were procedural rulings within the panel's discretion. Castellane had gotten a hearing, presented an expert, cross-examined Halloran Ridge's, and briefed the issue fully. Not viable.

Manifest disregard. The circuit had not squarely resolved whether the doctrine survived Hall Street. Even assuming it did, the award would need to show that the panel knew the governing rule and refused to apply it. The award instead showed the panel engaging with the rule and reaching a conclusion Castellane thought wrong. Not viable.

Step four: the recommendation

Nkechi's memorandum to the client ran two pages. The recommendation: do not move to vacate.

Her reasoning: the strongest ground carried a 10 to 15 percent chance; a motion would cost $150,000 to $250,000 through decision; a loss would add post-award interest running at the contractual rate on $18.4 million for the eight to twelve months the motion would take; and Halloran Ridge would move to confirm regardless, so the motion would be litigated on Halloran Ridge's timetable with Castellane in the posture of the party resisting a mandatory statutory command under § 9.

She added one more consideration. Castellane had four other supply contracts with materially identical limitation clauses. A published district court opinion holding that a panel may disregard such a clause and that the disregard survives review would be worse for Castellane than the $3 million.

What Castellane did. It paid, negotiated a modest discount for prompt payment, and — the part that mattered — revised its form limitation clause to state expressly that the limitation applies "regardless of the form of action, the theory of liability, or any characterization of the breach as fundamental, material, or total," and to state that the arbitrator lacks authority to award damages exceeding the cap. The second sentence is the operative one: it converts a merits question into a scope-of-authority question, which is the one question § 10(a)(4) actually reaches.

The lesson

The vacatur analysis and the drafting lesson point the same direction. If you want the arbitrator's authority limited, limit the arbitrator's authority in the clause — expressly, as a grant of power rather than as a substantive rule. A clause that says "damages are capped at X" gives the arbitrator a contract term to interpret. A clause that says "the arbitrator shall have no authority to award damages exceeding X" gives a reviewing court something to work with under § 10(a)(4).

Confirmation mechanics

Confirmation is the part practitioners under-prepare, on the theory that it is ministerial. It usually is, and the exceptions are expensive.

The application is a motion, not a complaint. Section 6 provides that any application to the court under the FAA "shall be made and heard in the manner provided by law for the making and hearing of motions." Filing a confirmation proceeding as a plenary civil action with a complaint and summons is a common error; some courts overlook it, others do not.

What to file. Section 13 specifies the papers: the agreement, the award, and each notice, affidavit, or other paper used on the application, plus any order confirming, modifying, or correcting. Attach them all. The most common defect in a confirmation motion is an incomplete record — an award without the arbitration agreement, or an agreement without the provision authorizing entry of judgment.

The judgment-entry requirement. Section 9 conditions confirmation on the parties having "agreed that a judgment of the court shall be entered upon the award." Most institutional rules supply this by incorporation — the AAA and JAMS commercial rules both provide that judgment may be entered on the award — and courts generally find the requirement satisfied by adopting such rules. But where the agreement is bespoke and silent, the question is live. Include the sentence in the clause.

Where to file. Section 9 permits application to the court specified in the agreement, or if none is specified, to the court in the district where the award was made. Subject matter jurisdiction is a separate question governed by Badgerow.

Timing. One year under § 9 for domestic awards; three years under § 207 for awards falling under the New York Convention. Courts are split on whether the one-year period is a limitations bar or permissive, but the safe course is to treat it as a deadline.

Interest. The award may specify pre- and post-award interest. Post-judgment interest, once the award is confirmed, runs under 28 U.S.C. § 1961 at the federal rate — which may be substantially lower than a contractual rate. A prevailing party with a high contractual rate should think about whether prompt confirmation is actually in its interest, and a paying party should think about the same question from the other side.

Appeals

Section 16 governs appellate jurisdiction, and it is deliberately asymmetric: orders hostile to arbitration are immediately appealable, orders favorable to arbitration generally are not.

Immediately appealable:

An order refusing a stay under § 3.
An order denying a petition to compel under § 4.
An order denying confirmation.
An order granting, continuing, or modifying an injunction against arbitration.
An order vacating an award without directing a rehearing.
A final decision with respect to an arbitration.

Not immediately appealable, absent § 1292(b) certification:

An order granting a stay under § 3.
An order directing arbitration to proceed.
An order compelling arbitration.

One further wrinkle. Coinbase, Inc. v. Bielski, 599 U.S. 736 (2023) held that a district court must stay its proceedings while an interlocutory appeal from the denial of a motion to compel arbitration is pending. The practical effect is that a defendant who loses a motion to compel gets an automatic pause, which changes the settlement dynamics of the appeal considerably.

Standard of review. On appeal from a confirmation or vacatur ruling, legal conclusions are reviewed de novo and factual findings for clear error — but the underlying deference to the arbitrator is unchanged. The court of appeals reviews the district court's application of an extraordinarily deferential standard, deferentially. Two layers of deference is a difficult posture from which to win.

Foreign and nondomestic awards

Chapter 2 of the FAA implements the New York Convention, and it changes the analysis in ways that matter even for awards rendered in the United States.

Chapter 2 applies more broadly than "foreign award" suggests. Section 202 brings within the Convention any award arising out of a commercial legal relationship that is not entirely between U.S. citizens — and even an award entirely between U.S. citizens if the relationship "involves property located abroad, envisages performance or enforcement abroad, or has some other reasonable relation with one or more foreign states."

Federal jurisdiction is automatic. Unlike Chapter 1, § 203 confers original federal jurisdiction over Convention actions regardless of the amount in controversy or the citizenship of the parties, and § 205 permits removal at any time before trial. Badgerow's jurisdictional problem does not arise here.

The defenses are Article V's, not Section 10's. Section 207 directs the court to confirm "unless it finds one of the grounds for refusal or deferral of recognition or enforcement of the award specified in the said Convention." Article V's seven grounds — incapacity or invalid agreement, lack of notice or inability to present a case, award beyond the scope of submission, irregular composition or procedure, award not yet binding or set aside at the seat, non-arbitrable subject matter, and public policy — overlap with § 10 but are not identical.

The three-year clock. Section 207 gives three years to seek confirmation, not one.

Where § 10 still matters. For an award made in the United States that falls under the Convention, most circuits permit the losing party to seek vacatur under Chapter 1's § 10 in the courts of the seat, while enforcement elsewhere is governed by Article V. The interaction is genuinely complicated, and a party facing a nondomestic award rendered in the United States should analyze both tracks.

Common misconceptions

"We can agree to appellate review in the arbitration clause." Not under the FAA. Hall Street forecloses it. What you can do is agree to an arbitral appeal — an internal appellate panel under the AAA or JAMS optional appellate rules, which reviews for material errors of law or clearly erroneous factual findings. The result is still an arbitral award subject to the same narrow judicial review, but the error correction happens inside the process, where it is permitted. If a client wants a second look, this is the mechanism.

"The arbitrator ignored the statute of limitations, so the award is void." Limitations questions are for the arbitrator. Howsam v. Dean Witter Reynolds, Inc., 537 U.S. 79 (2002) treats time limits as procedural questions presumptively for the arbitrator, not gateway questions for the court.

"There were no findings, so we cannot tell whether the arbitrator interpreted the contract." A bare award is not vulnerable for its silence. Courts presume regularity, and the absence of reasoning is not evidence of the absence of reasoning.

"The award is against public policy." In the labor context there is a narrow public policy exception, drawn from Misco, requiring that the policy be "explicit, well defined, and dominant" and ascertained from laws and legal precedents rather than general considerations of supposed public interests. It is narrow. In commercial arbitration it is narrower still, and it does not permit review of the merits.

"We can raise vacatur grounds whenever the other side moves to confirm." In most circuits, not after three months.

"Arbitration is cheaper, so this is all low-stakes." Awards in commercial arbitration routinely run into eight and nine figures, and the review that follows is thinner than the review of a small-claims judgment. The stakes of getting the clause right — the arbitrator's authority, the award form, the seat, the appellate option — are proportional to the amount at risk, not to arbitration's reputation for informality.

What this means for drafting

Everything above points back to the clause. Five drafting decisions determine how much review is available and how enforceable the result will be.

One — limit authority, not just liability. State the caps and exclusions as limits on the arbitrator's power, not only as substantive terms.

Two — specify the award form. Reasoned award for most commercial matters. Findings of fact and conclusions of law where a downstream use is likely.

Three — include the judgment-entry sentence. "Judgment upon the award rendered by the arbitrator may be entered in any court having jurisdiction thereof." It costs nothing and satisfies § 9.

Four — designate the court. Section 9 permits the parties to specify the court for confirmation. After Badgerow, naming a state court of competent jurisdiction — or a federal court where diversity will exist — avoids a genuinely wasteful fight.

Five — decide about appellate arbitration. The optional appellate rules add time and cost. For a clause governing disputes that could be existential, they are usually worth it. For a clause governing routine vendor disputes, they are not.