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

A client forwards a lawsuit and says: we have an arbitration clause, get this into arbitration.

Before you can answer whether the dispute is arbitrable, you have to answer a prior question — who gets to decide that? — and that question has its own prior question, because whether the parties agreed to let an arbitrator decide arbitrability is itself something someone has to decide.

This sounds like a law school hypothetical designed to induce despair. It is instead the actual structure of American arbitration law, and it has real consequences. A party that wins the "who decides" question has usually won the case's forum, and forum determines a great deal about cost, timing, discovery, appeal rights, and the availability of class treatment.

The doctrine is layered but it is learnable. There are four layers, and they stack in a specific order.

Layer one: severability

The first principle is the oldest and the least intuitive: the arbitration clause is a separate agreement from the contract that contains it.

Prima Paint Corp. v. Flood & Conklin Manufacturing Co., 388 U.S. 395 (1967) established it. Prima Paint claimed the whole consulting agreement was induced by fraud — the consultant had concealed that it was about to file for bankruptcy. Prima Paint argued that a fraudulently induced contract could not send it to arbitration. The Supreme Court disagreed: a claim of fraud in the inducement of the contract generally goes to the arbitrator, while a claim of fraud in the inducement of the arbitration clause itself goes to the court.

Buckeye Check Cashing, Inc. v. Cardegna, 546 U.S. 440 (2006) extended this to a claim that the contract was illegal and void — deferred-presentment transactions alleged to be usurious under Florida law. The Court held the challenge went to the arbitrator, and stated the rule in three propositions:

  1. As a matter of substantive federal arbitration law, an arbitration provision is severable from the remainder of the contract.
  2. Unless the challenge is to the arbitration clause itself, the issue of the contract's validity is considered by the arbitrator in the first instance.
  3. This rule applies in state as well as federal courts.

What severability means practically. "The contract is void" is not an argument that keeps you out of arbitration. "The arbitration clause specifically was procured by fraud," or "I never signed anything," or "the person who signed had no authority to bind me" — those are different, and they are for the court.

The line that matters: a challenge to the existence of any agreement is for the court; a challenge to the validity of the container contract is for the arbitrator. Buckeye itself flagged the distinction, noting that its holding did not address whether an agreement was ever concluded.

Layer two: the default allocation

Assuming there is an agreement, which questions does a court decide and which does an arbitrator?

The organizing case is Howsam v. Dean Witter Reynolds, Inc., 537 U.S. 79 (2002), and its framework is a presumption running in two directions.

Questions of arbitrability are presumptively for the court. These are gateway matters about whether the parties are bound to arbitrate a particular kind of dispute:

  • Whether a valid arbitration agreement exists.
  • Whether the parties are bound by it.
  • Whether the agreement covers this controversy.

Questions of procedure are presumptively for the arbitrator. These are matters that "grow out of the dispute and bear on its final disposition":

  • Time limits, including statutes of limitations and contractual deadlines.
  • Notice requirements.
  • Conditions precedent such as negotiation or mediation periods.
  • Waiver, delay, and similar defenses to arbitrability.
  • Compliance with the institution's rules.

Howsam itself involved an NASD rule barring arbitration of claims more than six years old. Dean Witter wanted a court to enforce the six-year rule; the Court held it was a procedural question for the arbitrator, because such rules are "aptly to be answered by the arbitrator."

Two important refinements.

Conditions precedent are for the arbitrator. BG Group plc v. Republic of Argentina, 572 U.S. 25 (2014) applied the framework to a treaty requiring eighteen months of local litigation before arbitration. The Court treated the requirement as a procedural precondition for arbitrators, not a jurisdictional condition for courts. The reasoning travels beyond investment treaties to ordinary commercial escalation clauses.

Waiver by litigation conduct is for the court, and the test is not what most courts had said. Morgan v. Sundance, Inc., 596 U.S. 411 (2022) eliminated the prejudice requirement that eight circuits had grafted onto waiver analysis. A party that litigates for months and then moves to compel does not get a special arbitration-favoring rule; ordinary waiver principles apply, and prejudice is not required unless the state's general waiver law requires it. The Court's broader point matters: "The federal policy is about treating arbitration contracts like all others, not about fostering arbitration."

Layer three: delegation

The default allocation is only a default. Parties may agree to send arbitrability itself to the arbitrator, and when they do, courts enforce it.

First Options of Chicago, Inc. v. Kaplan, 514 U.S. 938 (1995) set the standard. Courts should not assume the parties agreed to arbitrate arbitrability "unless there is clear and unmistakable evidence that they did so." The reasoning is a candid statement about arbitration's unusual posture: a party who has not agreed to arbitrate a question should not have it decided by an arbitrator whose decision receives only the most deferential review, and the ordinary presumption in favor of arbitration is reversed on this specific issue because a party is unlikely to have thought about who should decide arbitrability.

Rent-A-Center, West, Inc. v. Jackson, 561 U.S. 63 (2010) made delegation clauses independently enforceable and turned severability into a two-level doctrine. The agreement contained a provision giving the arbitrator "exclusive authority to resolve any dispute relating to the ... enforceability ... of this Agreement." Jackson argued the whole agreement was unconscionable. The Court held that because a delegation provision is itself an agreement to arbitrate a gateway question, severability applies to it: unless Jackson specifically challenged the delegation provision, the unconscionability challenge to the agreement as a whole went to the arbitrator.

The practical lesson is a drafting-and-pleading lesson. A party resisting arbitration in the face of a delegation clause must challenge the delegation clause specifically, with arguments directed at that provision, not at the agreement generally. Briefs that attack the arbitration agreement as a whole lose on Rent-A-Center without the court reaching the merits.

Henry Schein, Inc. v. Archer & White Sales, Inc., 586 U.S. 63 (2019) removed one escape hatch. Several circuits had recognized a "wholly groundless" exception permitting a court to decide arbitrability, notwithstanding a delegation clause, where the argument for arbitrability was frivolous. The Court rejected it unanimously. "When the parties' contract delegates the arbitrability question to an arbitrator, the courts must respect the parties' decision as embodied in the contract." A court "possesses no power to decide the arbitrability issue," even if it thinks the argument is obviously wrong.

What Henry Schein did not decide is at least as important: whether incorporating institutional rules constitutes clear and unmistakable evidence of delegation. The Court expressly reserved it, and the question remains live.

Layer four: the incorporation question

Most arbitration clauses do not contain an express delegation provision. They say the dispute will be arbitrated under the AAA Commercial Rules, or the JAMS rules, or the ICC Rules — and those rules say the tribunal may rule on its own jurisdiction.

Does adopting such rules clearly and unmistakably delegate arbitrability?

Most federal circuits have said yes, at least between sophisticated commercial parties. The reasoning: the parties agreed to rules that expressly empower the arbitrator to decide jurisdiction, and agreeing to the rules is agreeing to their content.

But the answer is not uniform, and the qualifications matter.

Consumer and employment contracts. Several courts have held or suggested that incorporation by reference is not "clear and unmistakable" when one party is an unsophisticated consumer or employee who never saw the rules. The reasoning has force: a person clicking through terms of service has not meaningfully agreed that an arbitrator will decide whether the arbitration clause is unconscionable.

Which version of the rules. Rules change. Whether the version in effect at contracting or at filing applies can matter if the delegation language was added or amended.

Partial incorporation. A clause that adopts an institution's rules "as modified herein" and then modifies them raises a question about whether the delegation survived.

The safe drafting answer: if you want delegation, say so expressly. A sentence costs nothing and eliminates the argument.

The arbitrator, and not any federal, state, or local court or
agency, shall have exclusive authority to resolve any dispute
relating to the interpretation, applicability, enforceability,
scope, or formation of this arbitration agreement, including any
claim that all or any part of it is void or voidable.

And if you want courts to decide arbitrability, say that — expressly carving gateway questions out of the delegation. Sophisticated parties sometimes prefer this, particularly where the scope of the clause is complex or where multiple related agreements have different dispute resolution provisions.

When agreements conflict: Coinbase v. Suski

The layered framework assumes one agreement. Modern commercial life often produces several.

Coinbase, Inc. v. Suski, 602 U.S. 143 (2024) presented the problem cleanly. Users had accepted Coinbase's User Agreement, which contained an arbitration clause with a delegation provision. They later entered a sweepstakes governed by Official Rules containing a forum selection clause designating California courts for disputes related to the sweepstakes.

When the sweepstakes was challenged, Coinbase moved to compel arbitration under the User Agreement. The plaintiffs pointed to the Official Rules.

The Court held that a court, not an arbitrator, decides which agreement governs. The reasoning: before a court can determine whether the parties delegated arbitrability, it must determine which contract controls — and that antecedent question of contract formation and superseding agreements is for a court. A delegation clause in the first agreement cannot bootstrap the arbitrator into deciding whether a second agreement displaced the first.

The practical significance is large, because layered agreements are everywhere:

  • A master services agreement with arbitration, and a statement of work with litigation.
  • Terms of service with arbitration, and a promotional or beta program with different terms.
  • An employment agreement with arbitration, and a separate equity award agreement with a Delaware forum clause.
  • A purchase agreement with arbitration, and a transition services agreement with a different mechanism.
  • A settlement agreement that supersedes an underlying contract, and the question of which dispute resolution provision survives.

The drafting response is to address the relationship expressly. Every subsequent agreement between the same parties should say whether it supersedes, supplements, or is subordinate to the earlier dispute resolution provision. A sentence — "The dispute resolution provisions of the Master Agreement govern any dispute arising under this Statement of Work" — prevents the entire problem.

What is not arbitrable, and who says so

Two categories deserve separate treatment because they are exceptions to almost everything above.

Section 1 of the FAA exempts "contracts of employment of seamen, railroad employees, or any other class of workers engaged in foreign or interstate commerce" from the Act's coverage. This is not an arbitrability question in the ordinary sense — it is a question whether the FAA applies at all — and courts decide it.

The Supreme Court has returned to § 1 repeatedly:

The Ending Forced Arbitration of Sexual Assault and Sexual Harassment Act, enacted in 2022 and codified at 9 U.S.C. §§ 401–402, gives a person alleging sexual assault or sexual harassment the option to void a predispute arbitration agreement as to that dispute. The statute expressly assigns to a court, not an arbitrator, the determination whether it applies — "irrespective of whether any party contests the enforceability" of the agreement or "asserts a waiver, defense, or other challenge," and notwithstanding any delegation clause. This is one of the few express congressional overrides of the delegation framework.

Other constraints on arbitrability are narrower than parties often hope. The Court has held antitrust claims arbitrable in the international context in Mitsubishi Motors Corp. v. Soler Chrysler-Plymouth, Inc., 473 U.S. 614 (1985), securities claims arbitrable, RICO claims arbitrable, and statutory employment claims arbitrable. The remaining ground — that a claimant cannot effectively vindicate a federal statutory right in arbitration — was substantially narrowed by American Express Co. v. Italian Colors Restaurant, 570 U.S. 228 (2013), which held that the fact that individual arbitration made a claim economically irrational did not invalidate a class waiver.

Class and collective arbitration as a gateway question

Whether a clause permits class arbitration is itself contested territory, and it interacts with the delegation framework.

The availability of class arbitration is a substantive gateway question in most circuits, presumptively for the court, because the difference between bilateral and class arbitration is fundamental rather than procedural. The reasoning traces to Stolt-Nielsen S.A. v. AnimalFeeds International Corp., 559 U.S. 662 (2010), which emphasized that class arbitration changes the nature of the proceeding so significantly that it cannot be presumed from silence.

Silence does not authorize class arbitration. Stolt-Nielsen vacated an award imposing class arbitration where the parties had stipulated the agreement was silent.

Ambiguity does not either. Lamps Plus, Inc. v. Varela, 587 U.S. 176 (2019) held that an ambiguous agreement cannot provide the necessary contractual basis for class arbitration, and that the state-law contra proferentem rule could not be used to find consent to class procedures.

But if an arbitrator is properly presented with the question and interprets the contract, the interpretation stands. Oxford Health Plans LLC v. Sutter, 569 U.S. 564 (2013) affirmed an award imposing class arbitration because the arbitrator had interpreted the contract — even though the interpretation was, in the Court's words, potentially "good, bad, or ugly." The distinction from Stolt-Nielsen is that Oxford had submitted the question to the arbitrator.

And class waivers are enforceable. AT&T Mobility LLC v. Concepcion, 563 U.S. 333 (2011) held the FAA preempts state rules conditioning enforceability on the availability of class procedures; Epic Systems Corp. v. Lewis, 584 U.S. 497 (2018) held that the National Labor Relations Act does not override class waivers in employment arbitration agreements.

The practical upshot for drafting: if you do not want class arbitration, say so explicitly, and place the class waiver where a court will find it — not buried in incorporated rules. And decide, deliberately, whether the class waiver's enforceability is delegated or reserved to a court; both are defensible choices with different risk profiles.

The procedural mechanics

The doctrine determines the answer; the FAA determines how it is presented.

Section 4 — petitions to compel. A party aggrieved by another's refusal to arbitrate may petition a district court for an order compelling arbitration. The court, "upon being satisfied that the making of the agreement for arbitration or the failure to comply therewith is not in issue," shall order arbitration. If the making of the agreement is in issue, the court proceeds summarily to trial of that issue — a jury trial if a party demands one and the issue is jury-triable.

Section 3 — stays. Where a suit is brought on an issue referable to arbitration, the court "shall on application of one of the parties stay the trial of the action." Smith v. Spizzirri, 601 U.S. 472 (2024) resolved a circuit split by holding that a court must stay rather than dismiss when a party requests a stay. The word "shall" is mandatory, and the practical consequences are real: a stayed case stays on the docket, which preserves jurisdiction for later confirmation — a meaningful benefit after Badgerow v. Walters, 596 U.S. 1 (2022) removed look-through jurisdiction for confirmation petitions.

Jurisdiction for § 4 petitions. Vaden v. Discover Bank, 556 U.S. 49 (2009) permits a court to look through the petition to the underlying controversy. This works for § 4 and not for §§ 9 and 10.

Section 16 — appeals. The asymmetry is deliberate. An order denying a motion to compel is immediately appealable; an order granting one generally is not. Coinbase, Inc. v. Bielski, 599 U.S. 736 (2023) added that a district court must stay its proceedings while an interlocutory appeal from the denial of a motion to compel is pending. The combination gives a defendant that loses a motion to compel an automatic pause and an immediate appeal — a significant strategic asset.

A worked problem

Thackeray Analytics licenses a data platform to Corvinus Retail Group under a Master License Agreement signed in 2022. The MLA contains an AAA commercial arbitration clause with no express delegation provision. In 2024 the parties signed a Professional Services Addendum for an implementation project; the Addendum contains a Delaware forum selection clause and a jury waiver, and says nothing about the MLA's arbitration clause.

The implementation fails. Corvinus sues in Delaware Superior Court for breach of the Addendum, breach of the MLA, fraudulent inducement of the MLA, and unjust enrichment. Thackeray moves to compel arbitration.

Thackeray's counsel, Oluwaseun Ravensworth-Mbeki, works the layers in order.

Layer one — is there an agreement at all? Yes. Both parties signed the MLA; formation is not disputed. The fraudulent inducement claim attacks the MLA as a whole, not the arbitration clause specifically. Under Prima Paint and Buckeye, that challenge goes to the arbitrator.

Layer two — which agreement governs? This is the hard one, and it is Coinbase v. Suski. The Addendum's forum clause and the MLA's arbitration clause point different directions, and a court decides which controls. Oluwaseun cannot skip this step by pointing to delegation, because there is no delegation of the question which contract applies.

Her argument: the Addendum governs the implementation services and is subordinate to the MLA, which by its terms governs "this Agreement and all Addenda, Statements of Work, and Orders." The Addendum's forum clause addresses disputes "arising under this Addendum" and does not purport to supersede the MLA's dispute resolution provision.

Corvinus's argument: the Addendum is later in time, contains an integration clause, and its forum and jury-waiver provisions would be nullities if arbitration governed.

Layer three — if the MLA governs, who decides scope? The MLA has no express delegation clause but adopts the AAA Commercial Rules, which empower the arbitrator to rule on jurisdiction. Most circuits treat incorporation as clear and unmistakable evidence between sophisticated commercial parties, and both parties here are sophisticated. But the question is unsettled — Henry Schein expressly reserved it — and Delaware's treatment matters.

Layer four — the scope question itself. The MLA clause covers disputes "arising out of or relating to this Agreement." The MLA claims plainly fall inside. Whether the Addendum claims do is exactly the question the Suski analysis controls.

How Oluwaseun structures the motion. She leads with the contract-hierarchy argument, because it is the antecedent question and because she can win it on the documents. She argues delegation in the alternative, and she is careful to frame it as an alternative rather than as an answer to the Suski problem — a brief that treats delegation as dispositive of which contract governs invites a court to notice the error.

What Corvinus's counsel, Priya Ellingham-Nakashima, does. She attacks in the other order. First, the Addendum governs, so no arbitration. Second, if the MLA governs, incorporation of the AAA rules is not clear and unmistakable delegation as to claims arising under a different agreement. Third, on scope, the Addendum claims do not arise out of or relate to the MLA.

And she does one more thing that matters. She does not attack the arbitration clause generally in a way that would trigger Rent-A-Center. Every argument is aimed at the antecedent question or at the specific delegation, not at the MLA's validity.

The outcome. The court holds that the MLA governs claims under the MLA and the Addendum governs claims under the Addendum, compels arbitration of the MLA claims, and stays the Delaware action as to those claims under § 3 while retaining the Addendum claims. Both sides get half of what they wanted, which is the most common outcome when two agreements point different directions.

Three lessons.

One — the antecedent question comes first and cannot be delegated. Suski is now the starting point in any matter with more than one agreement.

Two — a party resisting arbitration must aim carefully. Attacking the agreement as a whole in the face of a delegation clause is the single most common error, and it loses without the court reaching the merits.

Three — this entire dispute was a drafting failure. One sentence in the Addendum — "Disputes arising under this Addendum shall be resolved as provided in Section 14 of the Master License Agreement" — would have avoided a nine-month, several-hundred-thousand-dollar fight about forum.

What good drafting looks like

Decide who decides, and say it. Express delegation, or express reservation. Do not rely on incorporated rules.

If you delegate, use language that survives a specific attack. Courts have enforced delegation provisions that identify themselves as separate agreements and that the parties acknowledge separately.

Delegation. The Parties agree that the arbitrator, and not any
court, shall have exclusive authority to resolve any dispute
relating to the interpretation, applicability, enforceability,
scope, or formation of this Section [__], including any claim
that all or any part of this Section is void or voidable. The
Parties acknowledge that this delegation provision is a
severable agreement, that it may be enforced independently of
the remainder of this Section and of this Agreement, and that a
challenge to any other provision does not affect it.

If you reserve arbitrability to courts, say that too.

Notwithstanding the incorporation of the [institution] Rules,
questions concerning the existence, scope, or validity of this
arbitration agreement shall be decided by a court of competent
jurisdiction and not by the arbitrator.

Address the relationship among agreements — every time. Any addendum, statement of work, order form, amendment, settlement, or side letter should state which dispute resolution provision governs it. This one habit prevents most Suski problems.

Draw the scope deliberately. "Arising out of or relating to" is broad and includes tort and statutory claims connected to the relationship. "Arising under" is narrow. Pick one on purpose.

Put the class waiver where it can be found, in the arbitration section, in its own subsection, and decide expressly whether its enforceability is delegated.

Address the § 1 exemption in agreements with drivers, couriers, delivery personnel, and anyone who moves goods. After Bissonnette, the employer's industry is not the test, and a company that does not think of itself as a transportation business may still have transportation workers. A state-law arbitration fallback — providing that if the FAA does not apply, the agreement is enforceable under a designated state arbitration act — is the standard response.

Address the EFAA by acknowledging in the agreement that claims within its scope may be brought in court at the claimant's election. This costs nothing and avoids an argument that the clause is overbroad.

Consider carve-outs on purpose. Many commercial clauses except claims for injunctive relief to protect intellectual property or confidential information. These are sensible, and they must be drafted so they do not swallow the clause — a carve-out for "any claim seeking injunctive relief" invites every claimant to add an injunction request.

Non-signatories

A recurring practical question is whether someone who did not sign the arbitration agreement can compel or be compelled.

The governing principle is that ordinary state-law contract and agency doctrines apply. The FAA does not displace them, and the Supreme Court has said so repeatedly — most recently in the international context in GE Energy Power Conversion France SAS v. Outokumpu Stainless USA, LLC, 590 U.S. 432 (2020), which held that the New York Convention does not conflict with domestic equitable estoppel doctrines permitting non-signatory enforcement.

The recognized theories:

  • Incorporation by reference. A signatory to one contract that incorporates another containing an arbitration clause may be bound.
  • Assumption. A non-signatory that conducts itself as bound may be held to it.
  • Agency. An agent may enforce a principal's clause, and a principal may be bound by an agent's.
  • Veil piercing and alter ego. Where corporate separateness is disregarded generally, it is disregarded here.
  • Estoppel. The most litigated. A party that sues on a contract cannot avoid the contract's arbitration clause; and in some formulations, a signatory may be compelled to arbitrate with a non-signatory where the claims are intertwined with the contract and with the non-signatory's relationship to a signatory.
  • Third-party beneficiary. A beneficiary that claims contract benefits takes the burdens too.

Who decides whether a non-signatory is bound? This is a formation question, and courts generally decide it — even in the face of a delegation clause — because a delegation clause in an agreement someone never joined cannot bind them. The reasoning parallels New Prime: a provision cannot supply authority over a party the agreement does not reach.

Practical drafting. In group structures and supply chains, identify who should be covered and say so. A definition of "Party" that includes affiliates, successors, and permitted assigns, plus an express statement that specified third parties may enforce the clause, is far more reliable than litigating estoppel.

Where courts still divide

Four questions remain genuinely open or split, and they are worth knowing because they determine where a fight is worth having.

One — incorporation of rules as delegation, outside the sophisticated commercial context. Henry Schein reserved the question. The trend in consumer and employment cases runs against finding delegation from incorporation alone, but it is not uniform, and state courts applying state contract law reach varied results.

Two — whether a delegation clause survives a challenge that the entire contract was never formed. Buckeye reserved the formation question, and courts have divided on hard cases: forged signatures, agents without authority, terms presented in ways that may not have created assent at all. The general answer is that formation goes to the court, but the boundary between formation and validity is contested.

Three — the scope of the "wholly groundless" problem after its rejection. Henry Schein eliminated the exception, but courts continue to face clauses whose delegation language is arguably narrower than the arbitrability question presented. The analysis has migrated from "is this groundless" to "what did the delegation actually cover," which is a better question but not an easier one.

Four — who decides class arbitrability when the clause delegates generally. Most courts treat class availability as a gateway question for the court absent clear delegation of that question specifically. Some read a general delegation clause to cover it. The difference matters enormously, and it is worth resolving in the clause.

Why the structure is what it is

It is worth stepping back to ask why American law makes this so layered, because the answer explains where the doctrine will go.

Arbitration is a creature of consent. The FAA's operative provision, § 2, makes written arbitration agreements "valid, irrevocable, and enforceable, save upon such grounds as exist at law or in equity for the revocation of any contract." The whole edifice rests on the parties having agreed.

That premise generates the layers. If arbitration rests on consent, then a court must confirm consent before compelling. But consent can extend to who decides consent's scope — parties can agree that an arbitrator resolves gateway questions — so courts must confirm that second-order consent too. And because a delegation provision is itself an agreement, the severability logic applies to it as well.

The result is recursive but principled: at every level, the question is what the parties actually agreed to, and a court decides that until the parties have clearly agreed that someone else should.

Morgan v. Sundance is the clearest recent statement of the underlying commitment. Courts may not invent arbitration-favoring rules that depart from ordinary contract principles. The federal policy, the Court said, "is about treating arbitration contracts like all others, not about fostering arbitration." Read alongside Coinbase v. Suski, the direction is toward ordinary contract analysis applied rigorously — which is, in the end, easier to advise on than a thumb on the scale.

For practitioners, the operational summary is short. Identify every agreement between the parties. Determine which governs the claim, in a court, first. Then ask whether that agreement clearly and unmistakably delegates arbitrability. If it does, compel and let the arbitrator sort out scope. If it does not, brief scope to the court. And if you are resisting, aim your challenge at the specific provision that stands between you and the courthouse — not at the agreement in general.

A one-page decision map

For the lawyer holding a complaint and an arbitration clause, the sequence is:

1. Is there more than one agreement between these parties? If yes → a court decides which governs. Coinbase v. Suski. Do this first.

2. Does the FAA apply? Check § 1 — transportation workers, per Saxon and Bissonnette. Check the EFAA for sexual assault and harassment claims. A court decides these, delegation notwithstanding. New Prime.

3. Is the challenge to formation or to validity? Formation ("I never agreed") → court. Validity ("the contract is void, unconscionable, illegal") → arbitrator, under Prima Paint and Buckeye.

4. Is there a delegation provision? Express provision → arbitrator decides arbitrability, and a challenger must attack the delegation specifically. Rent-A-Center; Henry Schein. Only incorporated rules → circuit- and context-dependent; strongest between sophisticated commercial parties. Neither → court decides arbitrability under First Options.

5. If the court decides, is the question substantive or procedural? Existence, coverage, who is bound → court. Timeliness, notice, conditions precedent, compliance with rules → arbitrator, under Howsam. Waiver by litigation conduct → court, with no prejudice requirement after Morgan v. Sundance.

6. Class or collective procedures? Presumptively for the court absent clear delegation of that question. Silence and ambiguity do not authorize class arbitration. Stolt-Nielsen; Lamps Plus.

7. Filing mechanics. Move to compel under § 4; look-through jurisdiction available under Vaden. Always request a stay under § 3 — it is mandatory on request after Spizzirri, and the retained case preserves jurisdiction for later confirmation. If denied, appeal immediately under § 16, and note that district court proceedings are stayed during the appeal under Coinbase v. Bielski.

Frequently asked questions

Our clause incorporates the AAA rules. Is that enough to delegate arbitrability? Probably, between sophisticated commercial parties, in most circuits. Not reliably in consumer or employment contexts. And Henry Schein expressly left the question open. If it matters, add an express delegation provision.

The other side says the whole contract is void. Do we still get arbitration? Yes, under Buckeye — unless the challenge is aimed at the arbitration clause itself, or at formation rather than validity.

We litigated for eight months before deciding to arbitrate. Is that waiver? Very possibly. Morgan v. Sundance removed the prejudice requirement, so the analysis is ordinary waiver law. Move to compel at the outset or accept the risk.

Can we appeal if the court denies our motion to compel? Yes, immediately, under § 16 — and district court proceedings are stayed during the appeal under Coinbase v. Bielski.

Should we ask for a stay or a dismissal? A stay, always. Spizzirri makes it mandatory on request, and the retained case is the cleanest route to federal confirmation later given Badgerow.

Our client is a delivery company that does not think of itself as a transportation business. Does § 1 matter? Yes. After Bissonnette, the employer's industry is not the test — what the workers do is. Include a state-law arbitration fallback.

We have a master agreement with arbitration and an SOW with a forum clause. Which wins? A court decides, under Coinbase v. Suski. And then fix the template so the next one says.

Related documents

This article is general information, not legal advice, and does not create an attorney-client relationship.