Limitation of Liability Clauses Explained: Caps, Carve-Outs, and When They Fail

A limitation of liability clause — usually labelled "Limitation of Liability", "Liability Cap", or occasionally just "Limitations" — is among the most financially consequential provisions in any commercial contract. It tells the parties in advance what the maximum amount either side can recover if the other breaches. A contract where the other party's breach has materially harmed you can still produce only the cap amount in damages, even if your actual loss was many times larger.

This article explains how these clauses are structured, the carve-outs that typically cannot be capped, and the statutory doctrines under UK, US, and Australian law that override a cap when it goes too far. It is a factual overview, not legal advice. The UK Unfair Contract Terms Act 1977 and Consumer Rights Act 2015, Section 2-719 of the Uniform Commercial Code in the United States, and the Australian Competition and Consumer Commission's consumer-guarantees guidance are each useful starting points for the respective jurisdiction.[¹][²][³][⁴]

What the cap actually limits

A typical limitation of liability clause has two moving parts:

A cap amount. Often expressed as the greater of a dollar (or pound) figure and a multiple of the fees paid under the contract. Common formulations include "the total fees paid in the twelve months preceding the claim" and "£1,000,000 or the fees paid in the prior year, whichever is higher."
A type-of-damages exclusion. The clause usually also excludes categories of loss entirely — "indirect", "consequential", "special", "incidental", or "punitive" damages — regardless of the cap.

Read these together. A clause that caps total liability at "fees paid in the prior year" but also excludes "any indirect or consequential damages, including loss of profits, loss of goodwill, and loss of data" can produce an outcome where the paying party has large direct losses, large consequential losses, and an upper bound that makes the contract very difficult to enforce in practice. If the contract is commercially meaningful, the cap should be at a level that actually compensates for realistic breach scenarios.

Aggregate versus per-claim

Another structural point that changes the economics:

Aggregate cap. The total maximum the breaching party pays across all claims, for the entire contract term. Favours the breaching party.
Per-claim cap. Each distinct claim is subject to its own cap. More favourable to the party suffering breach.
Annual cap. The cap resets each year. Less common, typically a middle-ground negotiation outcome.

A clause that says "the aggregate liability of the Supplier under this Agreement shall not exceed the fees paid in the twelve months preceding the first claim" is aggregate plus one-year look-back. The supplier has maximum protection. A clause that says "the aggregate liability of each party in any twelve-month period shall not exceed fees paid in that period" is mutual and annual — much more balanced.

Direct versus consequential damages

The distinction matters enormously:

Direct damages. Damages that flow naturally from the breach — the difference between what was paid and what was received, the cost of replacement, the lost bargain.
Consequential damages. Losses that flow indirectly — lost profits, business interruption, loss of reputation, damage to other systems or data as a downstream effect of the breach.

A clause that excludes "indirect, consequential, special, incidental, and punitive damages" often turns out to exclude most of the real-world loss. In an IT services contract, if the service fails and the customer's business loses revenue as a result, the lost revenue is usually characterised as consequential. The direct damages — the service fees the customer paid for the broken service — may be a small fraction of the lost revenue.

The boundary between direct and consequential is contested and jurisdiction-specific. Some jurisdictions treat certain lost profits as direct damages if they flow naturally from the breach and were foreseeable at contract formation (the Hadley v. Baxendale line of cases). Others are stricter. A contract that defines specific categories of loss (loss of data, loss of goodwill) as "consequential" and excludes them is trying to get ahead of the contested line — and often succeeds in doing so.

Carve-outs: what the cap cannot cover

Almost every well-drafted LoL clause has carve-outs — specific categories of loss that are NOT subject to the cap. Common carve-outs:

Fraud and wilful misconduct. Virtually universal. A party that commits fraud or intentional wrongdoing cannot rely on a liability cap in most common-law jurisdictions, regardless of what the contract says.
Death or personal injury caused by negligence. Under the UK Unfair Contract Terms Act 1977, a clause that purports to exclude or restrict liability for death or personal injury caused by negligence is void — no amount of contract drafting overrides this.[¹] Similar rules apply in other common-law jurisdictions.
Indemnification obligations. Indemnities are often carved out of the general LoL cap because they function as a separate risk-transfer mechanism. The specific indemnities — IP infringement, confidentiality breach, specific statutory violations — are the obligations most commonly uncapped.
IP infringement. Particularly in technology and licensing contracts. The IP indemnity is typically uncapped because the downstream losses to the indemnified party from a third-party infringement claim can be substantial.
Confidentiality. A breach of confidence can produce reputational and competitive damage that vastly exceeds the fees paid. Confidentiality-breach liability is commonly carved out or subject to a much higher cap.
Payment obligations. The fees the buyer owes for services or products received are usually excluded from the cap — a cap on the buyer's payment obligation would defeat the purpose of the contract.
Statutory consumer rights. In jurisdictions with strong consumer-protection legislation (UK Consumer Rights Act 2015, Australian Consumer Law), certain consumer rights cannot be excluded or limited by contract.[²][⁴]

Read the carve-outs carefully. A LoL clause with a tight cap but generous carve-outs can be a fair allocation of risk. A LoL clause with a tight cap and no meaningful carve-outs is one-sided.

Mutual versus one-sided

Is the cap mutual (applies to both parties) or only to one side?

Mutual. Both parties face the same cap on the same terms. Common in balanced commercial contracts.
One-sided. Only the supplier's or only the customer's liability is capped. Common in supplier-drafted agreements where the customer's liability is largely payment obligations (which are usually carved out anyway).
Asymmetric amounts. Mutual but at different levels — supplier capped at fees paid, customer at a higher fixed figure that covers reasonable supplier losses from misuse, data exfiltration, etc.

A one-sided cap is a red flag unless there is a specific commercial reason for it (e.g., a low-cost service where uncapped exposure would make the economics impossible).

When a cap fails to hold

Even when the contract cleanly caps liability, courts in common-law jurisdictions can refuse to enforce the cap:

UCTA and the reasonableness test (UK, B2B contracts). The Unfair Contract Terms Act 1977 subjects most exclusion and limitation clauses in UK business-to-business contracts to a reasonableness test. Factors include bargaining strength, whether the buyer had insurance, and whether the clause was one-sided.[¹] A clause that fails the reasonableness test is unenforceable to that extent.
Consumer Rights Act 2015 (UK, B2C contracts). In UK consumer contracts, specific statutory rights cannot be excluded at all, and unfair terms more broadly are not binding on the consumer.[²]
UCC §2-719 (US, sale of goods). Where a limited remedy "fails of its essential purpose" — typically because the remedy becomes unavailable or cannot make the buyer whole — the limitation fails too, and the buyer has access to the full menu of UCC remedies.[³] Separately, UCC §2-719 prohibits exclusion of consequential damages where such exclusion would be "unconscionable", which is presumed for personal-injury claims in consumer sales.
Australian Consumer Law (AU). Certain consumer guarantees apply regardless of the contract and cannot be excluded by limitation clauses in consumer transactions.[⁴] Similar protections exist for small-business contracts under the ACL unfair-terms regime.
Fundamental breach and repudiation. Some older English and Commonwealth authority treated "fundamental breach" as a route around LoL clauses, but this doctrine has been significantly narrowed by more recent authority. A LoL clause that clearly addresses the relevant breach generally holds even when the breach is serious — unless statute or the unreasonableness doctrine intervenes.

The practical implication: the cap often holds, but not always. A party relying entirely on the cap to sleep at night is taking on enforceability risk; a party who believes the other side's cap is unreasonable should not assume the contract will automatically enforce it.

Negotiation moves

When the cap is too tight:

Raise the floor. Fees paid in the prior year is often too low for a contract where the breach could cause substantial downstream loss. Negotiate a fixed minimum floor (whichever is higher between fees paid and a fixed sum).
Carve out specific high-impact scenarios. Data breaches, IP infringement, wilful breach, breach of confidentiality — each can be separately carved out of the cap without changing the overall structure.
Insurance-backed caps. In some industries, the cap can be pegged to the supplier's actual insurance coverage, which is often higher than a simple fees multiple.
Annual reset. An aggregate cap that never resets is tight; an aggregate cap that resets annually is more generous to the claimant.
Per-claim caps for high-frequency breaches. Services with many small touchpoints — ongoing consulting, data processing, SaaS — often benefit from per-claim caps because any single claim may not reach the aggregate cap.

Red flags

Cap at fees paid in the prior twelve months with no minimum floor. In a new contract, there may be no "prior twelve months" of fees; the cap rounds to zero.
Exclusion of consequential damages that lists categories broad enough to swallow direct damages. "Loss of data, loss of profits, loss of business, loss of goodwill" — in a data-processing contract, loss of data IS the direct damage.
Uncapped indemnification obligations on one side only. If the customer indemnifies the supplier uncapped but the supplier's liability to the customer is tightly capped, the economic allocation is asymmetric.
"Maximum aggregate liability regardless of the form of action or theory of liability." Standard language that tries to prevent a claimant from recharacterising a claim as negligence or breach of statutory duty to escape the cap. Common and generally enforceable, but worth noticing.
No carve-out for wilful misconduct or fraud. Sometimes simply missing rather than deliberately omitted. Point it out — most counterparties will agree to add the carve-out.

Reading an LoL clause is ultimately about understanding what happens when the contract goes wrong. The cap and the carve-outs define the worst case. If the worst case feels acceptable on both sides, the clause is probably fair; if it feels unbalanced, the clause is probably doing exactly what the drafting party intended — and is the place to negotiate hardest.