A performance bond is a written guarantee from a bank or insurer (the surety) that pays an employer a capped sum — typically 10% of the contract value — if the contractor fails to perform or becomes insolvent. In UK construction it is one of the most common forms of security, sitting alongside parent company guarantees and collateral warranties. How and when you can actually recover under it depends almost entirely on the precise wording of the bond.
In short: A performance bond is a guarantee — usually from a bank or surety — that pays the employer a defined sum (commonly around 10% of the contract value) if the contractor fails to perform or becomes insolvent. Most UK construction bonds are “on-default” (conditional) bonds that require proof of breach and loss, rather than “on-demand” bonds. Performance bonds typically sit alongside other security such as parent company guarantees and collateral warranties.
That wording is the difference between getting paid in days and litigating for years. This guide explains what a performance bond is, the crucial split between on-demand and conditional (default) bonds, how a bond is called, the narrow fraud exception, and what happens when a contractor goes insolvent — with a worked £4m example and the latest 2026 court authority.
- A performance bond is usually a conditional (default) bond for around 10% of the contract sum — the employer must normally prove breach and loss before the surety pays.
- On-demand bonds pay against a bare written demand; the courts will only restrain a bank from paying in cases of clear fraud (Edward Owen v Barclays).
- A bond differs from a collateral warranty (which gives a third party a direct contractual claim) and from a parent company guarantee (which leans on group covenant rather than a third-party surety).
- In standard JCT contracts, the bond, guarantee and warranty provisions sit together in Part 2 of the Contract Particulars and Section 7 — the form must match what the bond actually says.
- Contractor insolvency does not automatically trigger a bond — Ziggurat v HCC shows you need either breach or express insolvency wording.
- Bond expiry, calls and quantum frequently surface during final account disputes, so timing the call before the longstop date is critical.
- What we see in practice
- Get the bond reviewed before signing: the time to fix weak triggers, short expiry and self-certification clauses is at drafting, not at the point of a call.
What is a performance bond?
A performance bond (also called a performance guarantee or, loosely, performance security) is a three-party instrument used to protect an employer against a contractor failing to perform its obligations under a construction contract. The parties are the contractor (the party whose performance is secured), the employer or beneficiary (who can claim under the bond) and the surety or guarantor — usually a bank or a specialist insurer.
The most common trigger for a claim is contractor insolvency before the works are complete. Where that happens, the bond provides compensation, guaranteed by a third party, up to the bond amount — money the employer can use to fund the additional cost of bringing in a replacement contractor to finish the works and remedy defects. The obligation to provide a bond is set out in the tender documents and the construction contract; the contractor then procures the bond and the cost is reflected in its price.
Crucially, a performance bond is normally a guarantee, not an insurance policy in the contractor’s favour and not a fund the contractor can draw on. It is security held by the employer. The surety’s exposure is capped at the stated bond amount, and the instrument typically expires on a defined date or event — most commonly practical completion, the end of the defects liability (rectification) period, or a fixed longstop date.
On-demand vs conditional (default) bonds
Whether a bond is “on-demand” or conditional turns on its wording. In Wuhan Guoyu Logistics Group Co Ltd v Emporiki Bank of Greece SA [2012] EWCA Civ 1629 the Court of Appeal approved the presumption in Paget’s Law of Banking: where an instrument relates to an international transaction, is issued by a bank, undertakes to pay “on demand” and does not set out a guarantor’s defences, it will almost always be read as an on-demand bond. The label on the document matters far less than these features.
The single most important distinction in this area is between on-demand bonds and conditional (default or “see to it”) bonds. They look superficially similar but behave very differently when money is at stake.
An on-demand bond creates a primary, autonomous payment obligation. The surety must pay on receipt of a conforming written demand, irrespective of the rights and wrongs of the underlying dispute. The employer does not have to prove breach or loss at the point of demand; the merits are argued out later. On-demand bonds are common in international and energy projects and are sometimes required by overseas employers, but they are riskier for contractors and therefore more expensive (or simply unavailable to smaller firms).
A conditional bond — the standard form for UK building work — creates a secondary obligation that is co-extensive with the contractor’s own liability. Payment is not triggered by a bare demand: the employer must normally establish the contractor’s breach of the underlying contract and the resulting loss. Because the surety’s liability mirrors the contractor’s, the surety can usually rely on the contractor’s defences (no breach, set-off, limitation), and liability is capped, typically at 10% of the contract sum.
| Feature | On-demand bond | Conditional (default) bond |
|---|---|---|
| Nature of obligation | Primary, autonomous — independent of the underlying contract | Secondary — co-extensive with the contractor’s liability |
| Trigger for payment | Conforming written demand only | Proof of breach and (usually) ascertained loss |
| Surety’s defences | Very limited — essentially only fraud or a non-conforming demand | Can rely on the contractor’s defences, set-off and limitation |
| Typical use | International projects, advance payment, some sub-contracts | UK building and engineering works |
| Cost to contractor | Higher; may require greater collateral | Lower; standard market product |
| Court restraint on a call | Only on clear evidence of fraud | Possible where the contract clearly bars the call |
Whether a bond is on-demand or conditional is a question of construction (interpretation), not of its label. In Yuanda (UK) Co Ltd v Multiplex Construction Europe Ltd [2020] EWHC 468 (TCC), Fraser J held that a guarantee was a conditional performance bond, not an on-demand instrument, precisely because the word “demand” and its synonyms were entirely absent and payment depended on liability arising from a breach of the sub-contract. The presence or absence of demand-language is often decisive.
How much is a performance bond worth — and what does it cost?
The bond amount (the maximum the surety will ever pay) is set in the contract and is conventionally 10% of the contract value. On very large or high-risk projects the figure is occasionally higher, and on some it is lower; 10% is the market default but not a legal rule. The bond amount is a ceiling, not an automatic entitlement — under a conditional bond the employer recovers only its proven loss up to that cap.
The premium (the cost of procuring the bond) is separate and far smaller. Premiums are typically in the region of 0.5% to 2% of the contract value, depending on the contractor’s financial strength, the bond wording, the project risk and the surety’s view of the covenant. A financially strong contractor pays less; a weaker one may pay more and be asked for collateral or counter-indemnities. The contractor bears the premium, which it builds into its tender price.
Performance bonds vs parent company guarantees vs collateral warranties vs retention
A performance bond is only one of several security mechanisms an employer may demand. They are complementary, not interchangeable, and a well-advised employer often takes more than one.
| Security | Who gives it | What it protects against | Key limitation |
|---|---|---|---|
| Performance bond | Bank or insurer (surety) | Contractor non-performance/insolvency, up to the bond cap (usually 10%) | Capped; conditional bonds need proof of breach and loss |
| Parent company guarantee (PCG) | The contractor’s parent company | The subsidiary’s performance and financial obligations | Only as strong as the parent’s covenant; worthless if the group fails |
| Collateral warranty | Contractor, consultant or sub-contractor | Gives a third party (funder, buyer, tenant) a direct contractual claim | Not a payment instrument — it creates a right to sue, not a fund |
| Retention | Withheld from the contractor’s own payments | Defects and incomplete work; an immediate cash buffer | Ties up the contractor’s cash; limited size (commonly 3–5%) |
The practical contrasts matter. A PCG relies on the financial stability of the contractor’s group; it provides no protection if the parent is itself in difficulty, whereas a bond brings in an independent, regulated surety. A collateral warranty is fundamentally different in kind: it is a separate contract that hands a third party (such as a funder or future tenant) a direct cause of action against a party they did not originally contract with — it secures a right to sue, not a pot of money. Retention is the employer’s own cash, held back from payments, giving an instant buffer but no third-party covenant. Bonds, PCGs, warranties and retention each cover a different gap, and the JCT suite is structured to allow an employer to require several at once.
How and when is a performance bond called?
Calling a bond means making a formal demand on the surety in the exact form the bond requires. The mechanics depend on the type of bond.
- Check the trigger. For a conditional bond, identify the event the bond names — usually a breach of the construction contract by the contractor, and sometimes insolvency where the bond expressly says so.
- Establish and ascertain the loss. Conditional ABI-type bonds typically require the employer’s loss to be “established and ascertained” — quantified by agreement, adjudication, arbitration or the court — before the surety must pay.
- Issue a conforming demand before expiry. The demand must come from the correct beneficiary, be in writing, attach any required proof (a certificate, adjudicator’s decision or judgment) and arrive before the bond expires.
- Pay within the cap. The surety pays the established sum up to the bond amount; anything above the cap remains a claim against the contractor (or its estate).
The wording of the demand clause is decisive. Some modern bonds let the employer self-certify the contractor’s default; others require an adjudicator’s award or court judgment as conclusive evidence. The recent decision in CR Construction v Barclays Bank (TCC, 4 February 2026) is a sharp illustration: a bond on a £117m Manchester residential scheme let under an amended JCT 2016 Design and Build contract allowed the employer to demand payment on a certificate countersigned by the employer’s agent, and the court treated that certified sum as conclusive of the bank’s liability. Contractors who agree self-certification wording give away a great deal of protection.
The fraud exception and injunctions
The leading authority is Edward Owen Engineering Ltd v Barclays Bank International Ltd [1978] QB 159, in which the Court of Appeal held that the issuer of an on-demand bond must pay against a conforming demand regardless of the rights and wrongs of the underlying contract. The only firmly established exception is clear, established fraud of which the bank has notice, and the courts set a high bar before they will restrain payment by injunction.
Contractors often want to stop a bond being paid when they dispute the underlying claim. The courts are extremely reluctant to interfere — and the more autonomous the bond, the harder it is.
The foundational authority is Edward Owen Engineering Ltd v Barclays Bank International Ltd [1978] QB 159. Lord Denning MR held that a performance guarantee is an obligation to pay on demand within the terms of the guarantee, “irrespective of the rights and wrongs” of any dispute between the parties, subject to one exception: “The only exception is when there is a clear fraud of which the bank has had notice.” Mere allegation of fraud is not enough; the evidence must be clear, and the bank must have notice of it. This is a deliberately narrow gateway, designed to protect confidence in bonds as “the life-blood of commerce”.
Two refinements matter for construction:
- Restraining the bank vs restraining the beneficiary. An injunction against the bank issuing or paying a bond will, in practice, only be granted for clear fraud. An injunction against the employer (the beneficiary) calling the bond may be possible on a different basis — where there is a strong case that an express term of the underlying contract prohibits the call. In Simon Carves Ltd v Ensus UK Ltd [2011] EWHC 657 (TCC), the court restrained a call not on fraud but because the contract said the bond became null and void on the issue of an acceptance certificate, which had been issued.
- The 2026 position. CR Construction v Barclays Bank confirms the courts’ continuing restraint. The contractor argued the demand was defective, the bond had been discharged by accepted repudiation, and sums were disputed. The TCC refused the injunction: there was no fraud (fatal to restraining the bank), the demand was substantively valid, clause wording expressly preserved liability after termination, and the balance of convenience — including the policy concern of not undermining the bond market — favoured refusal. The contractor had also delayed in applying.
The ABI model form of guarantee bond
The Association of British Insurers (ABI) model form of guarantee bond is the industry-standard conditional bond used on UK construction projects, although it is very frequently amended. Its core mechanism obliges the guarantor to “satisfy and discharge the damages sustained by the employer” following a breach of the construction contract by the contractor, with the guarantor’s liability co-extensive with the contractor’s and capped at the bond amount.
The form’s most litigated phrase is the requirement that the employer’s loss be “established and ascertained“. Yuanda v Multiplex (above) examined exactly this: the court analysed what “established and ascertained” means and confirmed the bond was conditional, requiring the sub-contractor’s liability to be determined and quantified — taking into account sums due or to become due — before the surety must pay. Earlier authority confirmed that an adjudicator’s decision can be capable of “establishing and ascertaining” damages under an ABI-type bond, which matters because adjudication is the usual quick route to a determination in construction.
Because the standard ABI form is so often amended, you cannot assume any two “ABI bonds” behave alike. Common amendments tighten or loosen the trigger, add insolvency wording, shorten the expiry, change the demand mechanics or convert the instrument towards on-demand. Reviewing the actual deed — not the label — is essential.
Interaction with insolvency of the contractor
If the contractor enters administration or liquidation, a performance bond can become one of the employer’s most valuable remedies, because the claim lies against the surety rather than the insolvent estate. The general statutory framework for insolvency is set out in the Insolvency Act 1986, and the timing of any bond call needs to be checked carefully against the bond’s own expiry and longstop dates.
Contractor insolvency is the scenario performance bonds are most often bought to cover — yet it is also where conditional bonds can disappoint employers who have not read the wording carefully.
The trap is that insolvency is not, by itself, a breach of the construction contract. A conditional ABI-type bond that pays only on “breach” may therefore not respond to insolvency alone unless either (a) the bond expressly extends to debts payable on insolvency, or (b) the employer can identify an actual breach — for example, the contractor failing to pay a sum that fell due to the employer following insolvency and termination.
This is exactly what Ziggurat (Claremont Place) LLP v HCC International Insurance Company plc [2017] EWHC 3286 (TCC) decided. Ziggurat employed a contractor to build student accommodation in Newcastle under a JCT 2011 form; the contractor suspended work, the employer terminated and brought in others, and the contractor became insolvent via a CVA. The bond was a standard ABI form with a bespoke clause 2 stating that damages payable “shall include (without limitation) any debt or other sum payable to the Employer under the Contract following the insolvency of the Contractor”. The court held that because of that bespoke wording the insolvency was enough to trigger recovery — and that, in any event, the contractor was in breach by failing to pay the amount due on insolvency. The lesson is blunt: if you want a bond to respond to insolvency, say so expressly.
Expiry, triggers and longstop dates
A bond is only useful while it is alive. Expiry is governed by the deed, and the typical options are an event (practical completion or the end of the defects liability/rectification period) or a fixed calendar longstop date — or the earlier of several. From the employer’s perspective it is usually wise to keep the bond in place until the final certificate at the end of the rectification period. From the contractor’s perspective, the bond should ideally expire on the earliest sensible event so its contingent liability and bonding capacity are released.
The commentary on CR Construction underscores good drafting: where no call has been made, contractors should seek expiry on the earlier of contract termination, practical completion, the end of the defects liability period or a specific longstop date, and may wish to require an adjudicator’s award or court judgment (rather than employer self-certification) as proof of default and quantum. Demands made after expiry generally fail, so an employer must diarise the longstop date and call in good time if a problem is crystallising.
| Stage | What typically happens | Bond status |
|---|---|---|
| Contract award | Bond required by tender; contractor procures it; premium paid | Bond issued, on risk |
| During the works | Surety on risk for breach/insolvency up to the cap | Live |
| Practical completion | Many bonds reduce or expire here unless extended to the rectification period | Live or partially released, per wording |
| End of rectification period | Final certificate; defects made good | Usually expires |
| Longstop date | Hard expiry regardless of project status | Expired — no demand possible after this |
Worked example: a 10% bond on a £4m contract and contractor insolvency
Scenario. Northgate Developments (the employer) engages Larchwood Construction Ltd to build a £4,000,000 mixed-use block under an amended JCT Design and Build 2016 contract. Larchwood provides a conditional ABI-type performance bond from a surety for 10% of the contract sum = £400,000. The bond expressly includes debts payable following the contractor’s insolvency (a Ziggurat-style clause) and expires at the end of the 12-month rectification period.
What goes wrong. With the works around 70% complete, Larchwood enters administration and stops on site. Northgate validly terminates and engages a replacement contractor.
The numbers:
- Extra cost to complete the works with the replacement contractor: £520,000 above what would have been paid to Larchwood.
- Additional preliminaries, professional fees and delay-related cost: £90,000.
- Retention already held by Northgate: £120,000.
How the bond responds. Northgate’s total recoverable loss is £520,000 + £90,000 = £610,000, less the £120,000 retention it can apply = £490,000 net. Northgate establishes and ascertains this loss (here, through adjudication), then issues a conforming written demand before the bond expires. The surety pays up to the cap: £400,000. The remaining £90,000 is an unsecured claim in Larchwood’s administration, where Northgate may recover only pence in the pound.
The lesson. A 10% bond covered most — but not all — of the shortfall, and only because (i) it expressly responded to insolvency and (ii) Northgate ascertained its loss and demanded before expiry. Without the insolvency wording, recovery would have turned on proving a breach; without timely action, the demand could have failed for lateness.
What we see in practice
In our advisory work, performance bonds tend to disappoint employers for one of two reasons. The first is discovering, only when a claim is needed, that the bond is a conditional “default” bond requiring proof of breach and loss, rather than the on-demand instrument the employer assumed it had. The second is letting the bond lapse: many bonds carry an expiry or longstop date tied to practical completion, and a call made a day late is worth nothing. We also see demands rejected for minor non-compliance with the bond’s formalities. Reading the actual bond wording at the outset, rather than at the point of dispute, is what makes the difference.
Common mistakes with performance bonds
- Assuming “10%” means automatic payment. Under a conditional bond the cap is a ceiling; the employer recovers only proven, ascertained loss.
- Relying on the bond for insolvency without insolvency wording. Insolvency is not a breach — without a Ziggurat-style clause the bond may not respond.
- Missing the expiry/longstop date. A demand made after expiry generally fails, however strong the underlying claim.
- Agreeing self-certification (contractors) or accepting weak proof clauses (employers). The proof mechanism decides how easily money moves — negotiate it deliberately.
- Trying to injunct a bank without fraud. Outside clear fraud, courts will not restrain a bank from paying (CR Construction 2026).
- Treating an ABI bond as standard. The form is “almost always amended” — read the actual deed, not the label.
- Confusing a bond with a guarantee or warranty. Each secures something different; one is rarely a substitute for another.
How Hayhills can help
Hayhills Legal Advisory provides direct, non-reserved advisory support on construction security. We review and draft performance bonds, parent company guarantees and collateral warranties, align them with your JCT or NEC contract, and pressure-test the trigger, proof, expiry and insolvency wording before you sign — the stage where value is won or lost. We also advise on calling or resisting a bond, ascertaining loss through adjudication, and protecting your position in final account disputes. Where a matter requires court enforcement, injunctions or insolvency litigation, we advise on strategy and introduce and coordinate a regulated solicitor to conduct the reserved steps, so you keep one clear line of advice throughout.
Need a performance bond reviewed or a call handled? Speak to the Hayhills construction advisory team on 0203 581 5789 or get in touch for a focused review before you commit.
Frequently asked questions
What is a performance bond in UK construction?
It is a written guarantee from a bank or insurer (the surety) that pays the employer a capped sum — usually 10% of the contract value — if the contractor fails to perform or becomes insolvent. It is security held by the employer, not a fund the contractor can draw on.
What is the difference between an on-demand and a conditional bond?
An on-demand bond pays against a conforming written demand alone, regardless of the underlying dispute. A conditional (default) bond pays only once the employer establishes the contractor’s breach and ascertains its loss, and the surety can rely on the contractor’s defences. UK building work usually uses conditional bonds.
How much is a performance bond — is it always 10%?
The bond amount is conventionally 10% of the contract value, but that is a market default, not a legal rule; it can be higher or lower. The premium to procure it is separate and much smaller — typically around 0.5% to 2% of contract value, depending on the contractor’s covenant and the wording.
How is a performance bond called?
You make a formal written demand on the surety in the exact form the bond requires, from the correct beneficiary, before expiry. For a conditional bond you must usually first establish and ascertain your loss — by agreement, adjudication, arbitration or court — and attach any required proof such as a certificate or award.
Can a contractor stop a bond being paid?
Rarely. A court will only restrain a bank from paying on clear evidence of fraud (Edward Owen v Barclays). It may restrain the employer from calling where an express contract term clearly bars the call (Simon Carves v Ensus), but in 2026 CR Construction v Barclays reaffirmed the courts’ strong reluctance to intervene.
What is the fraud exception?
It is the narrow rule that a court will only interfere with payment under a performance bond where there is clear fraud of which the bank has notice. A bare allegation is not enough; the evidence must be clear. The exception protects confidence in bonds as the “life-blood of commerce”.
Does a performance bond cover contractor insolvency?
Not automatically. Insolvency is not itself a breach of contract, so a conditional bond that pays only on “breach” may not respond unless it expressly covers insolvency debts (as in Ziggurat v HCC) or you can show an actual breach, such as non-payment of a sum due on termination.
What is the ABI model form of guarantee bond?
It is the industry-standard conditional bond produced by the Association of British Insurers, widely used on UK projects but very often amended. The guarantor pays the employer’s damages following the contractor’s breach, capped at the bond amount, once the loss is “established and ascertained”.
How does a performance bond differ from a collateral warranty?
A bond is a payment instrument: a third-party surety pays a capped sum on the contractor’s default. A collateral warranty is a separate contract giving a third party (such as a funder or tenant) a direct right to sue a contractor or consultant — it creates a cause of action, not a pot of money.
When does a performance bond expire?
On the event or date in the deed — commonly practical completion, the end of the defects liability/rectification period, or a fixed longstop date, sometimes the earliest of several. Demands after expiry generally fail, so employers should diarise the longstop date and call in good time if a problem crystallises.
This article is for general information only and does not constitute legal or accountancy advice. Hayhills Limited, trading as Hayhills Legal Advisory, provides non-reserved legal advisory services. Always check current requirements at GOV.UK.