The question usually arrives after the event. A heads of agreement has been signed, one party has changed its mind, and the other wants to know whether the document can be enforced. The honest answer is that some can and some cannot, and the difference has nothing to do with what the document calls itself.
Australian law has a settled framework for answering the question, and it is older than most people expect. It comes from a dispute over a farm sale in Western Australia, decided by the High Court in 1954.
Masters v Cameron and the three classes
In Masters v Cameron (1954) 91 CLR 353, the parties signed a memorandum for the sale of a farming property. The document said the agreement was made subject to the preparation of a formal contract of sale acceptable to the vendor's solicitors. The purchasers ran into financial difficulty and wanted out. The vendor said a contract already existed. The High Court said it did not.
Along the way, the Court described three classes of preliminary agreement. In the first, the parties have finalised their bargain and intend to be bound immediately, while proposing to restate the terms later in a fuller or more precise form. In the second, the parties have agreed all the terms and intend no departure from them, but have made performance of some obligation conditional on the execution of a formal document. In the third, the parties do not intend to be bound at all unless and until a formal contract is signed.
Agreements in the first two classes are binding contracts. An agreement in the third class binds no one. The memorandum in Masters v Cameron itself fell into the third class, largely because of the "subject to" language, and the purchasers were free to walk away.
The fourth class
The three classes did not stay three for long. The High Court had already recognised, in Sinclair Scott & Co Ltd v Naughton (1929) 43 CLR 310, a further kind of case: one where the parties are content to be bound immediately and exclusively by the terms they have agreed, while expecting to make a further contract in substitution for the first, containing additional terms by consent.
The New South Wales Court of Appeal applied that idea in GR Securities Pty Ltd v Baulkham Hills Private Hospital Pty Ltd (1986) 40 NSWLR 631. There the correspondence spoke of a "legally binding agreement in principle". The court held the parties were bound at once, even though a fuller contract was still to come. The words "legally binding" were given their ordinary force, and "in principle" did not undo them.
The practical significance of the fourth class is considerable. It means a heads of agreement can be a complete, enforceable contract today even though everyone at the table expects a longer document next month. The later document, if it ever arrives, replaces the first. If it never arrives, the first still stands.
What the courts actually look at
The classes are a way of organising the answer, not a test in themselves. The real question in every case is whether the parties intended to be immediately bound, and that intention is assessed objectively, from what the document says and what the parties did, not from what either of them privately believed.
The language of the document carries the most weight. "Subject to contract" and its variants have a long judicial history, and the usual effect of those words is that no contract exists until the formal document is exchanged. That presumption can be displaced, but a party arguing against the natural meaning of "subject to contract" starts a long way behind.
Conduct matters too. If both parties began performing after signing, paying money, handing over keys, commencing work, a court will find it easier to conclude they regarded themselves as bound. Silence and delay tend to cut the other way.
Then there is certainty. Even where the intention to be bound is clear, a document that leaves essential terms unresolved may fail as a contract because there is nothing sufficiently complete to enforce. Price, parties, subject matter and any term the parties themselves treated as essential must be settled or ascertainable. A court can fill small gaps by implication. It cannot write the bargain for the parties.
An agreement to agree is not an agreement
This is where many heads of agreement quietly fail. A clause that says the parties "will negotiate the remaining terms in good faith" or "will agree a price mechanism" is, on traditional analysis, an agreement to agree, and an agreement to agree is not enforceable, because a court cannot know what the parties would eventually have agreed.
The line between a binding contract with formal documents to follow and a mere agreement to agree can be fine. A document that fixes all the commercial terms and simply contemplates lawyers tidying it up sits on the enforceable side. A document that records the shape of a deal while leaving real commercial points open sits on the other. Plenty of heads of agreement sit awkwardly in between, which is exactly why they generate litigation.
Part performance and estoppel
Even where no contract was formed, a party who walks away is not always free of consequences. If one side has acted to its detriment on the assumption that a binding deal existed or would be completed, and the other side created or encouraged that assumption, an estoppel may arise. The High Court's decision in Waltons Stores (Interstate) Ltd v Maher (1988) 164 CLR 387 grew out of exactly this territory: lease negotiations that were never formally completed, and a landowner who demolished and started building in reliance on the deal proceeding.
Part performance plays a related role where land is involved, and acts done under an informal agreement can in some circumstances make it enforceable despite the absence of the usual formalities. Neither doctrine is a substitute for a properly formed contract. Both are expensive to run and uncertain in outcome. They are the arguments you reach for when the drafting has already failed.
Words that bind, words that do not
If the parties want the heads of agreement to bind, the drafting should say so plainly: this document constitutes a binding agreement between the parties, effective on signing. If they want the opposite, "subject to contract" remains the most reliable formula in the Australian cases, ideally reinforced by an express statement that no party intends to be legally bound until formal documents are exchanged.
The dangerous documents are the ones that do neither, or that try to do both at once. A heads of agreement that is expressed to be non-binding but contains obligations drafted in mandatory language invites a court to pick it apart clause by clause. As we explained in our overview of heads of agreement, particular provisions such as confidentiality and exclusivity are often intended to bind even within an otherwise non-binding document, and the cleanest drafting says exactly which clauses fall on which side of the line.
When the other side walks away
If a counterparty abandons a signed heads of agreement, the first task is classification. Read the document against the Masters v Cameron classes and ask the objective question: did these parties, judged by their words and conduct, intend to be bound now?
If the answer is yes, the walking party is in breach, and the remedies are the ordinary contractual ones, damages or, where the subject matter justifies it, specific performance. If the answer is no, the inquiry turns to whether any individual clauses were binding on their own terms, and then to estoppel and reliance-based arguments. Timing matters in these disputes. Positions taken in early correspondence, before advice, have a way of resurfacing later, so it pays to characterise the document carefully before asserting anything about it.
None of this is a reason to avoid heads of agreement. They do useful work. It is a reason to decide, before signing, which class you want to be in, and to say so in words a court will not have to guess at.