Arbitration vs Court Litigation in India: How to Choose the Right Forum for a Commercial Dispute

Start with the statutory architecture, because it explains why the drafting-stage decision is so consequential. Section 7 of the Arbitration and Conciliation Act 1996 requires only that an arbitration agreement be in writing — a clause in the contract, an exchange of letters or emails, even an unsigned document the parties acted on can qualify. Once such an agreement exists, Section 8 makes judicial reference mandatory: a party sued in court on an arbitrable claim applies for reference, and the court must refer unless it finds prima facie that no valid arbitration agreement exists. The examination is deliberately shallow. In Vidya Drolia v. Durga Trading Corporation, (2021) 2 SCC 1, the Supreme Court confined the referral court to a prima facie review, sending all serious contests over validity and arbitrability to the tribunal itself.

The Court has since closed the most-litigated escape hatch. In In Re: Interplay Between Arbitration Agreements and the Stamp Act, a seven-judge bench held in December 2023 that an unstamped or insufficiently stamped agreement is not void — stamping is a curable defect, and objections on that ground do not belong at the Section 8 or Section 11 stage. Combined with the kompetenz-kompetenz principle — the tribunal rules on its own jurisdiction first — the practical position is this: if your contract contains anything resembling a written arbitration agreement, expect to arbitrate. Filing a suit in the teeth of a clause usually buys a few months of delay, an adverse costs order, and a reference anyway.

The tilt has a reach beyond signatories. In Cox and Kings Ltd v. SAP India Pvt Ltd (2023), a Constitution Bench affirmed the group-of-companies doctrine: a non-signatory affiliate can be bound by an arbitration agreement where its conduct — negotiating, performing, or being the intended beneficiary of the contract — shows a mutual intention to bind it. For corporate groups that route contracts through one entity while performance flows through another, this cuts both ways: it can drag a reluctant parent into your arbitration, and it can drag yours into someone else's. The lesson at the drafting table is that the arbitration clause is not boilerplate. It is a forum-selection decision with near-irreversible consequences, and it deserves the same attention as the price and liability clauses.

Not every dispute is arbitrable, and knowing the boundary saves wasted proceedings on both sides. Vidya Drolia laid down a four-fold test: a dispute is not arbitrable when it concerns actions in rem that bind the world rather than just the parties, when it affects third-party rights and requires centralised adjudication, when it relates to the inalienable sovereign or public-interest functions of the State, or when a statute expressly or by necessary implication reserves it for a specific forum. Applying that framework, the Court held that tenancy disputes governed by rent-control statutes are non-arbitrable, while ordinary lease and licence disputes under the Transfer of Property Act can be arbitrated. Insolvency, criminal offences, matrimonial status, guardianship, and testamentary matters stay with courts and tribunals.

The most commercially important clarification concerned fraud. For years, defendants resisted arbitration by alleging fraud and citing older authority that serious fraud belonged in open court. Vidya Drolia overruled that line: allegations of fraud between the parties are arbitrable, however serious, unless the fraud vitiates the arbitration agreement itself or raises issues against the State or with clear public overtones. The practical consequence is that pleading fraud no longer works as a forum-shifting device — a partner accused of siphoning funds, a distributor accused of fabricating invoices, a promoter accused of misrepresentation in a share-purchase agreement will all ordinarily face the tribunal, not a civil judge.

For a business at dispute time, the arbitrability screen is the first question, asked honestly: is the relief you actually need something a private tribunal can grant? A tribunal can award damages, declare contractual rights between the parties, and order specific performance. It cannot wind up a company, cannot adjudicate an insolvency, cannot cancel a registered instrument against the world, cannot punish anyone, and cannot bind a stranger to the agreement (Cox and Kings' non-signatory doctrine aside, which requires a real connection to the contract). If the endgame requires any of those, the arbitration clause is not your path even if it exists — and Section 3 of this analysis, on when court is genuinely better, becomes the operative chapter.

Arbitration's headline advantage is the statutory clock. Section 29A requires a domestic tribunal to make its award within twelve months of completion of pleadings, extendable by six months with the parties' consent and thereafter only by court order — with the court empowered to substitute arbitrators and dock fees for delay. A documents-heavy commercial dispute before a competent tribunal realistically produces an award in eighteen to thirty months from invocation, including the appointment stage. The proceedings sit on the parties' calendar, not a court's cause list: hearings run on consecutive days, adjournment culture is weaker, and there is no risk of the matter being passed over because a part-heard criminal appeal consumed the day.

The courtroom comparison must be equally honest in both directions. A contested commercial suit — even under the disciplined Commercial Courts Act regime of 120-day written statements, mandatory disclosure, and case-management hearings — realistically runs three to six years to a trial-court decree, and the first appeal adds years more. Before the suit is even instituted, Section 12A requires pre-institution mediation unless the plaintiff seeks urgent interim relief; the Supreme Court in Patil Automation Pvt Ltd v. Rakheja Engineers Pvt Ltd, (2022) 10 SCC 1, held the requirement mandatory and upheld rejection of non-compliant plaints. But the court track has a genuine fast lane arbitration lacks: Order XIII-A summary judgment lets a plaintiff with clean documents and a defendant with no real prospect of defending obtain a decree without trial — there is no equivalent of a summary award in Indian arbitration practice.

The number that deflates the arbitration timeline is the challenge tail. An award is not the end: the losing party has three months (extendable by thirty days) to apply under Section 34 to set it aside, and a Section 34 decision is appealable under Section 37. The grounds are deliberately narrow — Ssangyong Engineering & Construction Co Ltd v. NHAI (2019) confirmed that after the 2015 amendment there is no merits review, no reappreciation of evidence, and "patent illegality" applies only to domestic awards — so most challenges fail. But they fail slowly: a contested Section 34 petition and Section 37 appeal in a busy High Court can consume two to four years, during which enforcement continues only if no stay is granted. The realistic end-to-end comparison is therefore not "eighteen months versus five years" but "eighteen to thirty months plus a challenge tail, versus four to seven years plus appeals" — still an advantage, but a narrower one than the brochure version.

The structural cost difference is simple: in court, the State supplies the judge, the courtroom, and the registry, and the parties pay court fees — ad valorem on the claim value under state-specific schedules, capped in some states — plus their own professional engagement. In arbitration, the parties fund the entire adjudicative apparatus: the arbitrators' fees, the venue, transcription, and any administering institution's charges. None of this refers to advocates' fees, which in both forums depend entirely on the engagement the party negotiates with its own counsel; the comparison here is about the forum's own cost structure, which is where the tracks genuinely diverge.

For ad hoc domestic arbitration, the Fourth Schedule to the 1996 Act supplies the reference fee model that courts apply when they appoint arbitrators and that many tribunals adopt: a slab structure that scales with the sum in dispute and tapers at the top, with a ceiling on the sole arbitrator's fee and a premium for a sole arbitrator over a co-arbitrator's share. Two design decisions multiply or divide this cost head dramatically. A three-member tribunal costs roughly three times a sole arbitrator, which is defensible for high-stakes disputes and ruinous for modest ones. And institutional arbitration — under bodies such as the MCIA, DIAC, or IIAC — replaces open-ended fee negotiation with a published schedule, adds case-management discipline, and typically proves more predictable than ad hoc even where the headline numbers look similar.

The honest synthesis: for small claims, arbitration's fixed tribunal costs are regressive — the forum can cost a disproportionate share of the amount in dispute, which is why a low-value claim with clean documents is often better in a commercial court, where Order XIII-A summary judgment and modest capped court fees do the work. For mid-size and large claims, the tribunal cost is a minor percentage of the stakes and is routinely outweighed by the value of two to four years saved and by costs orders: arbitral tribunals in India award costs against the loser more realistically than civil courts historically have, and Section 31A's costs-follow-the-event regime applies to both tracks on paper but bites harder in arbitration practice. Budget for the whole journey, though — the Section 34/37 tail is fought in court, on court timelines, with court-side costs.

The old fear that choosing arbitration meant surrendering the power to freeze assets is obsolete. Section 9 of the 1996 Act lets a party apply to court for interim measures — securing the amount in dispute, restraining asset disposal or encashment of bank guarantees, appointing receivers, preserving goods — before the arbitration begins, during it, and even after the award but before enforcement. If a Section 9 order is obtained before the tribunal exists, the arbitration must be commenced within ninety days. Once the tribunal is constituted, Section 17 gives it the same interim powers, and a Section 17 order is enforceable as if it were an order of the court; Section 9 thereafter recedes unless the tribunal's remedy would be inefficacious. In urgency terms, a well-drafted Section 9 application before a commercial court moves as fast as an interlocutory injunction application in a suit — the protection is at parity.

Enforcement is where arbitration has quietly become stronger than its reputation. Under Section 36, an arbitral award is enforced in the same manner as a decree of the court, and — critically — the 2015 amendment abolished the automatic stay: merely filing a Section 34 challenge no longer suspends enforcement. The award-debtor must separately apply for a stay, and courts routinely condition any stay on deposit of all or a substantial part of the awarded amount, which transforms post-award settlement dynamics. Where an allegation of fraud or corruption in the making of the award is prima facie made out, an unconditional stay is possible, but that is the exception. A decree from a suit, by contrast, faces a first appeal as of right in which stays of execution remain commonplace and deposit conditions are less uniform.

Two further enforcement dimensions matter to specific businesses. Cross-border: India is a New York Convention jurisdiction, and a foreign award from a notified convention country is enforceable under Part II of the 1996 Act through a High Court, with resistance confined to narrow grounds — while an Indian court judgment is directly executable abroad only in a short list of reciprocating territories. Any contract with a foreign counterparty should therefore lean arbitration for enforceability alone. Confidentiality is the other quiet asymmetry: Section 42A obliges the arbitrator, institution, and parties to keep proceedings confidential except where disclosure is necessary for enforcement, whereas a suit's pleadings, orders, and judgments are public documents. For disputes touching pricing, technology, customer lists, or simply a listed company's reputation, privacy is not a soft benefit — it is often the decisive one.

Multi-party reality is the biggest one. Arbitration binds only those within the agreement's reach; a tribunal cannot implead your counterparty's sister concern that holds the assets, the bank that issued the guarantee, the sub-contractor whose default caused the breach, or the individual director you say gave the personal assurance — unless each is a signatory or squarely caught by the Cox and Kings non-signatory doctrine, which demands a demonstrable intention to be bound. A civil suit joins all necessary and proper parties in one proceeding under one judgment. If your dispute's real architecture involves three or more unaligned interests, forcing it through a bilateral arbitration produces fragmented proceedings, inconsistent findings, and multiplied cost — court is structurally the better forum.

Several other fact patterns point the same way. Where the counterparty's conduct is genuinely criminal — forged documents, siphoned funds meriting prosecution — the arbitration can proceed (Vidya Drolia makes inter-party fraud arbitrable), but only a court-side strategy integrates the civil claim with the FIR, and the criminal overlay often supplies settlement pressure no tribunal can. Where you need a precedent — a distributor terminated on a clause that binds your whole network wants a published, binding ruling, which a confidential award can never be. Where the counterparty is sliding into insolvency, the forum question dissolves entirely: IBC proceedings are non-arbitrable, a Section 14 moratorium will freeze the arbitration like any other proceeding, and the strategic analysis in our guide on NCLT insolvency versus commercial-court recovery takes over. And where the claim is small and documentary, the commercial court's summary-judgment track beats funding a tribunal.

There is also a statutory hybrid worth knowing: a registered micro or small enterprise with an unpaid buyer can file under Section 18 of the MSMED Act 2006 before the MSME Facilitation Council (the Samadhaan route), which attempts conciliation and, failing that, itself arbitrates or refers the dispute to institutional arbitration under the 1996 Act — statutorily, and regardless of whether the contract contains any arbitration clause. The award carries the Act's penal compound interest on delayed payments, and a buyer challenging it under Section 34 must first deposit seventy-five per cent of the awarded amount. For qualifying suppliers, this council arbitration is frequently the highest-leverage forum available, and it should be evaluated before either a suit or a contractual arbitration is launched.

Most Indian arbitration pathologies are drafting failures, and the most common is the appointment mechanism. In Perkins Eastman Architects DPC v. HSCC (India) Ltd (2019), the Supreme Court held that a person with an interest in the outcome of the dispute cannot unilaterally appoint the sole arbitrator — killing the classic clause under which one party's managing director names the tribunal. Clauses of that design now guarantee a fight: the counterparty ignores the unilateral appointment, applies under Section 11 to the High Court (or, for international commercial arbitration, the Supreme Court), and months are lost before a hearing on the merits is even possible. Draft a neutral mechanism from the start — appointment by a named institution, or by agreement with an institutional fallback — and the Section 11 detour never happens.

The seat is the second load-bearing choice. Since Bharat Aluminium Co v. Kaiser Aluminium Technical Services (BALCO, 2012), the seat determines which courts supervise the arbitration — Section 9 interim relief, Section 11 appointments, Section 34 challenges all go to the courts of the seat, regardless of where hearings physically happen or where the parties sit. Choose a seat whose commercial bench you are content to litigate the challenge tail in; for NCR parties a Delhi seat keeps supervision in a court system with deep arbitration experience. Name the seat expressly and separately from the venue, specify the governing law of the contract and, in cross-border deals, the law of the arbitration agreement itself — silence on these is what multi-year jurisdictional battles are made of.

The remaining checklist items each earn their sentence. Sole arbitrator versus three: default to a sole arbitrator unless the stakes clearly justify tripling the tribunal cost head. Institutional versus ad hoc: name a functioning institution — its rules will handle appointment, fees, timelines, and emergency relief, and you inherit none of the coordination failures of ad hoc practice. Language of the proceedings, if any counterparty or document set is non-English. Scope: "all disputes arising out of or in connection with this agreement" — resist carve-outs that invite parallel proceedings. Survival and severability, so termination of the contract does not take the clause down with it. And stamp the contract properly anyway: In Re: Interplay means an unstamped agreement is arbitrable, but curing stamp defects mid-dispute is delay you gifted yourself at the drafting table.

At the drafting table. Put the clause in when the relationship is bilateral, the likely disputes are commercial and inter-party, the counterparty is of comparable sophistication, the contract value justifies a tribunal, confidentiality matters, or any party is foreign. Leave it out — deliberately, not by oversight — when foreseeable disputes will involve third parties (guarantors, group companies, banks), when your realistic claims will be small and documentary so that the summary-judgment track is your best weapon, or when you are the party likelier to be defending and see no reason to fund a faster process against yourself. If you contract with MSME suppliers, price in the fact that the Samadhaan council route can override your dispute clause regardless of what you draft.

At dispute time, with a clause. Accept early that you will almost certainly arbitrate: Section 8's mandate, the prima facie standard from Vidya Drolia, and In Re: Interplay's burial of the stamping objection mean the reference is close to inevitable, and money spent resisting it is usually wasted. Redirect that energy into sequence: send the invocation notice; move under Section 9 immediately if assets may dissipate; push appointment fast — by the clause's mechanism if Perkins-compliant, by Section 11 application if not; then run the Section 29A clock hard. At dispute time without a clause, the court track applies: comply with Section 12A pre-institution mediation (or plead genuine urgency for interim relief), treat the mediation window as a real settlement opportunity rather than a formality, and build the plaint for an Order XIII-A summary-judgment strike if the documents are clean. A post-dispute arbitration agreement under Section 7 remains legally possible — in practice, a counterparty enjoying the courtroom's delays rarely signs one, but it is worth one letter.

The forum decision rewards being made deliberately, early, and with the whole journey in view — award or decree, challenge or appeal, and enforcement at the end of it. If you are weighing an arbitration clause in a contract on your desk, or sitting on a live dispute and unsure whether the clause in your agreement helps or hurts you, NyaySevak offers a free case assessment: share the contract and the dispute's outline, and we match you with experienced arbitration and commercial-litigation advocates in Delhi, Noida, Gurgaon, and other cities who can evaluate the forum question, the interim-relief position, and the realistic timeline before the first notice goes out. The first notice, as with most dispute strategy, often locks in the track — it is worth getting right.