Change order releases and reservation of rights

The mechanism is banal, which is part of the problem. A contractor is handed forty change orders over two years, prices each of them at its direct cost, signs each of them, gets paid, and moves on. Each carries a paragraph of standard language about complete compensation. Nobody negotiates it, because on any single change it is worth a few thousand dollars and the money is needed this month. At the end of the job the contractor assembles the cumulative impact claim it always intended to bring, and discovers it has been released forty times over.

The rest of the variation claim machinery is about establishing entitlement. This page is about keeping it. Two citations on this page are corrections to widely repeated errors, and both run in the direction opposite to the way commentary usually reports them.

What does standard change order release language actually do?

Standard change order release language converts a priced change order into a settlement of everything arising out of that change, including heads of loss nobody priced. Typical bilateral text provides that the adjustment "includes all costs associated with the scope of work… including all direct, indirect, and impact costs… such as loss of productivity, ripple effect, cumulative impact", and that it constitutes "compensation in full… for all costs and markup directly or indirectly… for all delays related thereto".

The regulatory source in US federal work is FAR 43.204(c), which directs that supplemental agreements should include release language stating that the contractor accepts the modification as a "complete equitable adjustment". That is why the boilerplate appears on the face of most bilateral SF 30 modifications: it is not an aggressive owner's drafting, it is the default the regulation asks for.

Signing at direct cost is therefore not the neutral act contractors treat it as. Pricing a change at what it costs to build the changed work, and signing an instrument that says the price is compensation in full for direct, indirect and impact costs, is an election. It says that the direct cost is the whole cost.

Two defences run off that signature, and they are not the same defence. Holland v United States, 621 F.3d 1366, 1377 (Fed. Cir. 2010) keeps them apart, and so should any pleading.

• Release is a contractual relinquishment, interpreted like any other contract term, with the focus on the parties' intent at the time of execution: Korte-Fusco Joint Venture, ASBCA No. 59767 (5 November 2015). The parallel citation the corpus previously carried, 15-1 BCA ¶ 36,158, does not sit with a November 2015 decision date and is not reproduced until it has been re-checked. If the words are unambiguous, extrinsic evidence about what the parties meant does not get looked at.
• Accord and satisfaction has four elements and the government or owner bears the burden as an affirmative defence: (1) proper subject matter, (2) competent parties, (3) a meeting of the minds, (4) consideration, Brock & Blevins Co. v United States, 343 F.2d 951, 955 (Ct. Cl. 1965). Whiting-Turner Contracting Co., ASBCA No. 56319, 10-1 BCA ¶ 34,436 (2010) adds that there must be "mutual agreement between the parties with the intention clearly stated and known to the contractor".

Meeting of the minds is the usual battleground, Meridian Engineering Co. v United States, 885 F.3d 1351 (Fed. Cir. 2018). O'Connor v United States, 308 F.3d 1233 (Fed. Cir. 2002). It is also the reason the two defences must be pleaded separately: an argument about what the parties actually understood is available on accord and satisfaction and may be shut out on a release whose words the tribunal finds unambiguous. That is exactly what happened in Bell BCI.

Bell BCI is the appellate reversal, and the contractor lost

Bell BCI Co. v United States, 570 F.3d 1337 (Fed. Cir. 2009) is the leading US authority on change order releases, and the Federal Circuit held for the Government. Any commentary describing Bell BCI as a case in which a cumulative impact claim was awarded despite release language is quoting the decision below: the Court of Federal Claims, 81 Fed. Cl. 617 (2008), which the Federal Circuit reversed. The two citations are routinely conflated, and the conflation reverses the outcome.

The facts are worth carrying because they are the reason the case is quoted for a contractor-friendly proposition it does not support. On an NIH building project the Government added a floor. The parties executed Modification 93, roughly $2.1m in consideration, providing that it constituted full compensation for the changed work and that the Contractor "hereby releases the Government from any and all liability under the Contract for further equitable adjustment attributable to the Modification". NIH then issued 113 further modifications carrying 2,016 extra work orders. The Court of Federal Claims awarded roughly $6.2m, including $2m of cumulative impact labour inefficiency and $1.6m of delay.

The Federal Circuit reversed. Paragraph 8's release was unambiguous, it barred the cumulative impact claims, and the court did not examine extrinsic evidence. The sentence that carries the whole doctrine is this one: "If parties intend to leave some things open and unsettled, their intent so to do should be made manifest." Judge Newman dissented, pointing to unchallenged trial findings that the parties never intended Modification 93 to release cumulative impact claims arising from work orders issued afterwards. The dissent is the fair reading of what the parties meant. It is not the law: the great majority of the changes came after the release, and the majority held the release caught them anyway.

Bell BCI remains controlling, and release language continues to receive ordinary contract interpretation, most recently in Honeywell International Inc. v General Services Administration, a 2024 CBCA decision whose docket number is not recorded in the sources consulted for this article and which should be located before it is cited in a pleading. The practical lesson is not that release language is unfair. It is that it is a contract term, read like any other contract term, and the burden of making an exception manifest sits on the party who wants the exception.

The reservation language that works

A reservation works when it names the claim, the event and the head of loss, sits inside the executed instrument or a document that instrument incorporates, and is repeated on every single change order. A general "all rights reserved" footer is materially weaker than a reservation that identifies what is being preserved, because the tribunal's question is not whether the contractor was unhappy. It is whether the parties made manifest what they were leaving open.

The best modern drafting model is the reservation upheld in Alares Construction, Inc. v Department of Veterans Affairs, CBCA 6149, 7071, 7597 (21 March 2025). Modification P00005, dated 23 May 2017, was a bilateral modification adding a second steam generator for $42,398.86, direct costs only. It incorporated RFIs 26 and 31, by which the contractor reserved its rights to:

"additional costs for impact of this change, alone or in combination with other changes, on unchanged work, for additional time, due to impacts, if any, on the schedule, and for time-related extended time of performance costs, all of which will be evaluated separately."

Three drafting devices are doing the work there, and each answers a specific defence. "Alone or in combination with other changes" pre-empts the argument that a per-change release disposes of an aggregate claim. "On unchanged work" identifies cumulative impact as the reserved head rather than leaving it to inference. "All of which will be evaluated separately" states the bifurcation expressly, so the instrument fixes the direct cost and nothing else.

The case law splits cleanly along that line, and the split is about drafting rather than doctrine.

Jurisdictional, United States only. Boston Shipyard (1st Cir.) and Vanlar (Cal. App.) are appellate. The board, Court of Federal Claims, district court and state trial decisions are all first instance.

Two of those are worth separating out because they are the recent ones and they cut in the contractor's favour. Thalle (13 August 2025) confirms that release language which is not clear is not a release, and that a mod-specific reservation of a named head (there, adverse weather) is given effect. Group III Management (2 March 2026) is narrower but useful: the modification addressed the drainage-revision work, not the procedural delay in funding and issuing the modification, and the Contracting Officer's own letter had described the issuing delay as "a separate issue". A release covers what the instrument addressed. It does not automatically cover the owner's conduct in getting the instrument out.

The knowing-waiver argument is available and it is not strong. Where the cumulative effect was not knowable when each modification was signed there can be no knowing and voluntary waiver, L.K. Comstock & Co. v Becon Construction Co., 932 F. Supp. 906, 937–38 (E.D. Ky. 1993). It is the same argument that runs against cardinal-change waiver. But Kleinknecht barred the claims even though damages were not yet calculable at execution, so do not build a strategy on it.

Alares upheld the reservation, and the claim still failed

Alares Construction, Inc. v Department of Veterans Affairs, CBCA 6149, 7071, 7597 (21 March 2025) is not a release case, and it should not be cited as authority that a release defeated a claim. The Board upheld the reservation of rights incorporated into modification P00005 and found no implicit release. The claim then failed largely on concurrency, an entirely independent ground. Commentary describing Alares as a case where the reservation "did not survive" inverts the holding.

What the Board actually did was bifurcate. The direct costs of the change were fixed by modification P00005 at $42,398.86. The time-related extended general conditions were preserved for separate evaluation, exactly as the reservation said they would be. The Department of Veterans Affairs argued that by executing P00005 the contractor had assumed responsibility for the extra work and the resulting delays. The Board rejected that and gave the reservation effect.

Then the claim met the schedule evidence. Of 765 days claimed, the Board found the VA responsible for 218. On the permanent-power design-defect issue it allowed 34 days, being the window in which permanent power was the sole critical path driver. That is a concurrency outcome, not a release outcome, and the analysis that decided it belongs to the concurrent delay body of law rather than to anything in this article.

A good reservation preserves a claim. It does not prove it.

One qualification on Alares's weight. The Board decided the reservation point on plain contract interpretation without citing published precedent on releases or accord and satisfaction. Its value is as a drafting model (the wording worked, against a sophisticated federal owner, in 2025) rather than as a doctrinal authority on release law. Cite it for the clause, not for the rule.

Cumulative impact is the claim being released

The head of loss that change order releases actually destroy is cumulative impact: the synergistic productivity loss on unchanged work caused by the aggregate of many changes, which by definition cannot be priced change by change. AACE International Recommended Practice No. 130R-23, Demonstrating Entitlement to Cumulative Impact Claims in Construction (15 May 2023) defines it as the "net impact of two or more undifferentiated changes… being much greater than the sum of the effect of the individual parts", resulting in reduced productivity of unchanged work.

Keeping three things apart decides whether a reservation can save anything.

Cumulative impact is not a global claim, and collapsing the first into the second is how a good claim gets struck. RP 130R-23 gives cumulative impact a recognised structure and a recognised place as a head of claim: it applies across all delivery methods and contract types, it is drafted for use where the contract has no procedure of its own, and its §2.2 entitlement elements run in order: §2.2.1 causal event occurrence, §2.2.2 adherence to contract change notice requirements, §2.2.3 demonstration of contractual entitlement, §2.2.4 demonstration of causation, §2.2.5 prudent mitigation, §2.2.7 quantification of impacts, with §2.3 treating document retention as a standalone requirement rather than an afterthought. Whether a §2.2.6 exists is unclear from the published table of contents.

Note where §2.2.2 sits. AACE puts notice and reservation compliance inside entitlement, not inside quantum, which is the right place for it, and which tells you the order to build the claim in. The release problem is a gating issue, argued before anybody looks at a productivity curve.

A global claim, by contrast, is a pleading of last resort: an undifferentiated assertion that the whole overrun was the owner's fault, advanced because the causal links were never recorded. It is not a head of claim with a structure. The RP is explicit on the related confusion, and it is the one a cross-examiner reaches for first: cumulative impact is a condition. The total cost method is a quantification technique. They are routinely muddled in expert reports. The credibility hierarchy for quantification runs measured mile first, then earned value or baseline comparison, then comparable project studies, then industry studies such as the MCAA productivity literature, and only then modified total cost and total cost.

The US textual hook is FAR 52.243-4(d), which extends the equitable adjustment to an increase in cost or time "of any part of the work under this contract, whether or not changed by any such order". Quote it in any US-facing cumulative impact argument, because it answers the instinctive objection that the contractor is claiming for work nobody changed.

Prolongation and thickened preliminaries are a different claim again, with their own trigger, notice and proof. The mechanics are in prolongation and delay costs. Running them together with cumulative impact in a single undifferentiated number is the fastest way to lose both.

Getting paid before the price is agreed

The reason contractors sign is cash, and the answer to the release problem is therefore partly a payment answer: take the money you are entitled to without signing for the money you are not. Every major form obliges the contractor to proceed with a directed change before the price is agreed, and three regimes now expressly separate the undisputed value from the disputed remainder so that the undisputed part is payable on the ordinary cycle.

• CCDC 2 – 2020, GC 6.3.11: "Pending determination of the final amount of a Change Directive, the undisputed value of the Work performed as the result of a Change Directive is eligible to be included in progress payments." The Contractor must proceed promptly under GC 6.3.4, so the payment valve is the quid pro quo. GC 6.3.11 itself is verified. The sub-paragraph numbering within the surrounding cost provisions at GC 6.3.6 and GC 6.3.7 is not, so cite the rule and check the decimals against the printed form before pleading them.
• AIA A201–2017, §7.3.9: pending final determination of the total cost of a Construction Change Directive, the Contractor may request payment for Work completed under the CCD in Applications for Payment. Use it rather than waiting for definitisation. Under AIA A201–2017 §7.3.6 the Contractor must promptly proceed on receipt of a CCD and advise the Architect of agreement or disagreement with the method of adjustment. There is no right to withhold performance pending pricing agreement. The internal wording of §7.3.9 was not read verbatim for this article, so do not quote it, cite it.
• California Civil Code § 8850, added by SB 440 and effective 1 January 2026, imposes the same split by statute on private works. The § 8850 mechanism, its deadlines and its interaction with AIA A201–2017 are set out below.

The strategic consequence is a pricing discipline rather than a legal argument. Price the variation so that a defensible undisputed core is visible on the face of the submission, with the contested elements identified and separately valued rather than folded into a single number an owner can dispute in its entirety and pay at nil. That is a cash-flow decision, not a presentational one, and it is the same discipline that makes an account survive a fast forum. How the core itself is built (which rung of the pricing ladder applies and what a fair valuation must contain) is a separate question and it comes first.

A partially disputed Change Directive paid at nil is a contractual debt, and in Ontario it is now expressly adjudicable: O. Reg. 264/25, s. 19, in force 1 January 2026, makes adjudicable "payment under the contract, including in respect of a change order, whether approved or not, or a proposed change order". There is no US analogue to statutory adjudication, which is why the American answer had to arrive as a payment statute instead.

California SB 440, the first US change-order payment statute

California SB 440 was signed on 10 October 2025 and took effect on 1 January 2026, adding Civil Code § 8850, referred to in commentary as the Private Works Change Order Fair Payment Act. Private works only. It does not apply to public works, and it operates alongside rather than instead of the existing prompt-payment provisions. It is the first US statute to attach a penal interest rate and a statutory stop-work remedy specifically to change-order work.

The mechanism is a mandatory, deadline-driven process:

1. The contractor submits a detailed, documented written claim for extra work or schedule relief.
2. The owner must meet and confer within 30 days, identifying disputed and undisputed portions.
3. The owner has 10 days after the meeting to issue written confirmation. Failure means the claim is treated as disputed.
4. Undisputed portions must be paid within 60 days, failing which interest accrues at 2% per month, 24% per annum.
5. Disputed items must go to non-binding mediation before litigation or arbitration.
6. Stop-work right: the contractor may suspend work 40 days after issuing a stop-work notice if the owner ignores the timelines or refuses mediation.

Additional dispute-resolution procedures are permitted only if they do not delay or dilute the § 8850 protections. That anti-avoidance provision is what makes the interaction with the standard form a live question rather than a drafting preference.

Then the hard part, which nobody has yet had to answer in a decided case. SB 440 effectively legislates a CCD-like structure onto private California work (segregate undisputed from disputed within a fixed window, pay the undisputed portion) mirroring the logic of AIA A201–2017 §7.3.9 with statutory teeth. It must be reconciled against A201–2017 §7.3, §15.1.3 and §15.4. Two frictions are identifiable on the face of the instruments. The 21-day Claim bar in §15.1.3 sits awkwardly with a statutory timetable built on 30-, 10- and 60-day steps: a contractor working the § 8850 process in good faith can run past day 21 while doing so, which is the same forfeiture trap A201 already generates when parties negotiate a change proposal by email. And the SB 440 mediation condition precedent overlays §15.4, adding a statutory pre-condition to a form that already sequences its own dispute machinery.

Neither friction has been resolved. No decided case on § 8850 exists, and the first SB 440 disputes are the thing to watch in California private work. Until one lands, the defensive position is straightforward: run the § 8850 process and serve the A201 Claim notice on the A201 timetable anyway, in the form AIA A201–2017 §1.6.2 requires: by certified or registered mail, or by courier providing proof of delivery, because emailed Claim notice is non-compliant unless the parties have amended §1.6.2. Doing both costs a stamp.

California is also alone in this. No other US state legislation enacted between 2024 and August 2026 is directed specifically at change orders or at prompt payment for changed work: Texas commentary in the window reflects the existing Prompt Payment Act, not new legislation.

Where the release problem is not a US problem

No equivalent body of doctrine exists in English, Australian, Canadian or Singaporean law. There is no comparable line of authority on change order releases as a distinct claim-killer, and this article does not generalise the US position to those jurisdictions. What exists instead are the ordinary tools: contractual construction of whatever the settlement wording actually says, and accord and satisfaction at common law. Where a contractor outside the United States signs an instrument saying the sum is in full and final settlement of everything arising from a change, that is construed as a contract term, but there is no developed body of case law on the specific problem of serial change orders extinguishing an aggregate impact claim, and asserting one would be inventing it.

Two things do travel, and they are worth naming because they occupy the same commercial space from opposite directions.

Australia has the mirror-image risk. In Kaloriziko Pty Ltd v Calibre Construction Group Pty Ltd (No 2) [2025] NSWCA 259, allowing an appeal from [2025] NSWSC 593 and reducing the builder's claim by AUD 3,243,175, the New South Wales Court of Appeal held that "approved" variations are not necessarily final and may be reassessed and unwound where the contract characterises payments as being "on account" only. The first-instance judgment had held the opposite. So the Australian contractor's problem is not that its approved variations are too final. It is that they may not be final at all. The defensive drafting response runs the other way too: provide expressly that variation approvals and pricing are final and binding once issued, or separate variation assessments from the "on account" progress-payment language entirely.

And every form with a quotation procedure reproduces the release problem without using the word. Under NEC4 ECC, a compensation event quotation under clause 62 prices both the time and the cost consequences, and implementation under NEC4 ECC clause 66 is final. The forecast is not revisited when the outturn is known. A contractor that quotes only the direct work has settled the time and disruption consequences of that compensation event just as surely as a US contractor who signs an unreserved modification. JCT SBC 2024's Schedule 2 Variation Quotation and JCT DB 2024's Schedule 2 Supplemental Provision 2 work the same way: direct loss and expense and the effect on time are priced inside the quotation, and acceptance settles those heads for that Change. Both are optional and apply only if stated in the Contract Particulars. Note in passing that JCT DB has no "Schedule 2 Quotation". That instrument is SBC's, and DB's analogue is Supplemental Provision 2.

Which leaves the drafting lesson, which is the only part of this that is genuinely universal, and it is the Federal Circuit's own sentence read as an instruction rather than as a holding: if you intend to leave something open, make it manifest: in the instrument, naming the head of loss, every time.

Sobre este material El contenido de la biblioteca es información general sobre la práctica de las reclamaciones de construcción, no asesoramiento jurídico. El derecho a reclamar, los plazos y el procedimiento se rigen por su propio contrato y por el derecho del lugar donde se ejecuta la obra. Revisado el 7 de agosto de 2026 · Informar de una corrección