Non-Compete Clauses and Restraint of Trade in Australia: What Holds Up

Somewhere near the back of most Australian employment contracts sits a clause nobody reads until the day it matters. It says that for six or twelve months after leaving, the employee will not compete with the employer, will not approach its clients, and will not recruit its former colleagues — sometimes within one city, sometimes across the whole country. Nobody negotiates it on the way in. Everyone fights about it on the way out.

The strange thing about restraint of trade clauses is that, as a class, they are presumptively void — and yet Australian courts enforce them regularly. The law starts from the position that a contract restraining a person from earning a living is against public policy, then makes an exception where the restraint protects something worth protecting and goes no further than reasonably necessary to protect it. Nearly every restraint dispute is fought on that ground, and most of the drafting devices in modern contracts — the cascading clauses, the elaborate definitions, the choice of New South Wales law — exist to win it.

This guide explains how Australian courts actually test post-employment restraints, why New South Wales differs from every other state, how these disputes run in practice, and what the announced federal ban on non-competes — flagged in the 2025-26 Budget but not yet law — would change.

Void until proven reasonable

The starting point is more than a century old and has never been displaced: a restraint of trade is contrary to public policy and void unless it is reasonable — reasonable between the parties, and reasonable in the interests of the public. The classic statement comes from Nordenfelt v Maxim Nordenfelt Guns and Ammunition Co Ltd [1894] AC 535, and the High Court of Australia has applied the doctrine ever since — most famously in Buckley v Tutty (1971) 125 CLR 353, where it struck down the New South Wales Rugby League's player transfer rules as an unreasonable fetter on a footballer's freedom to earn a living.

Two consequences follow. First, the onus of showing the restraint is reasonable as between the parties sits on the employer seeking to enforce it, not on the employee. Second, courts scrutinise employment restraints more strictly than restraints given on the sale of a business. A purchaser who pays for goodwill is entitled to protect what it bought; an employee bargaining for a job rarely negotiates from equal strength, and the courts know it.

What counts as a legitimate interest

An employer cannot use a restraint simply to keep a capable operator out of the market. Freedom from competition as such is not a legitimate interest — competition is the very thing public policy protects. The interests the cases recognise are narrower:

Customer connections. Where the employee was the human face of the business — the adviser clients call, the account manager who knows their pricing and their preferences — the employer has a real interest in a period of protection while a replacement builds the relationship. The closer and more personal the connection, the stronger the interest.
Confidential information. Genuine trade secrets and commercially sensitive information — pricing models, margins, product roadmaps, client strategies — can justify a restraint, on the practical logic that once an employee sits inside a competitor it is impossible to police what they remember and use.
Staff stability. Australian courts have accepted that an employer has a legitimate interest in maintaining a stable, trained workforce, which is what non-poaching covenants protect.

What the employer cannot claim is the employee's own general skill, experience and professional know-how. Those belong to the employee, however expensively they were acquired on the employer's time. A restraint whose real purpose is to sterilise the employee's ordinary skills — rather than to protect connections, secrets or the team — fails at this first hurdle, whatever its drafting says.

The four covenants — and why they are not equal

Most restraint clauses bundle several distinct promises, and courts treat them very differently:

Non-competition: a promise not to work for a competitor, or carry on a competing business, at all. The bluntest instrument, and the hardest to justify.
Non-solicitation of clients: a promise not to approach the employer's customers. Narrower, and far more readily enforced.
Non-dealing: a promise not to act for those customers even if they approach the employee unprompted. Broader than non-solicitation, but still tied to the customer connection the law protects.
Non-poaching of staff: a promise not to recruit former colleagues.

The practical rule of thumb: a court will ask whether the narrower covenants would adequately protect the employer's interest. A blanket non-compete generally survives only where non-solicitation and confidentiality obligations are not enough — classically, where the employee carries confidential information so woven into their working knowledge that compliance could never be monitored. If an employer could protect itself by a non-solicit, a non-compete restraining all work in the industry is vulnerable.

Reasonableness is measured at the date of the contract

A point that surprises both sides: reasonableness is assessed as at the date the restraint was agreed, in the circumstances then existing and reasonably foreseeable — not at the date of the departure. The High Court confirmed the principle in Amoco Australia Pty Ltd v Rocca Bros Motor Engineering Co Pty Ltd (1973) 133 CLR 288. So a restraint signed by a junior analyst is tested against what was reasonable for a junior analyst, even if the person leaving eight years later is a division head with the client book in their pocket. Employers who never refreshed the contract on promotion routinely discover their restraint was sized for a job the employee no longer holds. Conversely, a restraint imposed mid-career, in exchange for a promotion or pay rise, is tested against that senior role.

Within that frame, courts weigh three dials:

Duration. The yardstick is how long the employer reasonably needs — for client relationships to be re-anchored to a new contact, or for confidential information to go stale. For most employees that is measured in months, not years. Twelve months is defensible for genuinely senior, client-facing roles; for ordinary staff, even six can be a stretch.
Area. Geographic limits must match the footprint of the interest. In Lindner v Murdock's Garage (1950) 83 CLR 628, the High Court held unenforceable a restraint covering both towns in which the employer traded, when the employee had worked, and built connections, in only one. A nationwide restraint for an employee who served one city's clients is the modern version of the same mistake — though for online and national businesses, activity-based limits increasingly matter more than maps.
Activity. The restrained activities must correspond to what the employee actually did. Restraining a salesperson from holding any role of any kind with any competitor — including one with no customer contact — overreaches.

Cascading clauses and the blue pencil

Because an unreasonable restraint is void, and because courts will not rewrite a bad clause into a good one, drafters respond with the cascading or ladder clause: the restraint is expressed as a series of separate combinations — twelve, then six, then three months; Australia, then the state, then the city — with a provision that each combination is a distinct covenant. The idea is that a court can strike out the combinations that go too far and enforce the widest one that survives.

The common law tool for this is severance, often called the blue pencil test: a court may delete words or whole severable promises, but it may only cross out, never write in. If the reasonable and unreasonable parts cannot be separated by deletion alone, the whole restraint falls. Cascading clauses are drafted precisely to make deletion easy — but they have limits. A clause generating hundreds of permutations invites an argument that it is uncertain or oppressive, and a court asked to choose among dozens of alternatives may conclude the employer itself never knew what protection it genuinely needed.

The New South Wales difference

One state changed the rules. Under the Restraints of Trade Act 1976 (NSW), a restraint is valid to the extent to which it is not against public policy. In practical terms, a New South Wales court can read an over-wide restraint down and enforce it to the extent that is reasonable — rather than striking it out because it was drawn too broadly. An employer whose clause says two years and Australia may still get an injunction for six months and Sydney, if that narrower protection is justified on the facts.

This makes New South Wales the most enforcement-friendly restraint jurisdiction in the country, and it is why so many national employment contracts select New South Wales governing law. The choice is not a magic wand — there must be a genuine connection to the state — but where it holds, the difference between reading down and striking out can decide the case. Everywhere else in Australia, the blue pencil is all there is.

How these disputes actually run

Restraint litigation is sprint, not marathon. The pattern repeats: the employee resigns; the employer learns — often from a LinkedIn announcement or a client's passing remark — where they are going; a letter arrives within days demanding written undertakings to honour the restraint; and if undertakings are refused, the employer applies urgently for an interlocutory injunction to hold the position until trial.

At that hearing the employer must show a serious question to be tried and that the balance of convenience favours restraint — and must give the usual undertaking as to damages, promising to compensate the employee if the injunction later proves to have been wrongly granted. That promise has real teeth: an employee kept out of work for months by an injunction that fails at trial has a claim, and our guide to the cross-undertaking in damages explains how it is enforced. In practice, the interlocutory hearing usually is the case. By the time a trial could be listed, the restraint period has expired; whoever wins the injunction fight tends to win the war, and most matters settle on undertakings shortly after.

Springboard relief

Where the real complaint is misused confidential information — the classic modern fact pattern is a forensic IT report showing client lists copied to a USB drive or a personal email account in the final week of employment — the employer may seek a springboard injunction. The logic is simple: a person who takes confidential material to get a head start in competition can be restrained for long enough to cancel that head start, even where the restraint clause itself is doubtful. Springboard relief runs alongside claims for damages or an account of profits for breach of contract and breach of confidence.

Garden leave and the restraint clock

Garden leave is the restraint's gentler cousin: the employee serves out a long notice period at home, on full pay, employed but idle — unable to work for anyone else because they are still employed. Because the employee is paid throughout, courts view garden leave more benignly than an unpaid restraint. But the two interact: what matters is the total time the employee is kept out of the market. Six months in the garden followed by a twelve-month restraint is eighteen months on the sidelines, and the restraint will be judged accordingly. Well-drafted contracts set garden leave off against the restraint period; where the contract does not, the employee can still run the point on reasonableness.

Evidence employers need — and defences employees run

For the employer

Injunctions are won on evidence, not indignation. The employer needs the signed contract containing the restraint — the current one, not a superseded version from two roles ago; concrete evidence of the interest, such as revenue by client and the employee's role in those relationships; evidence of breach or imminent breach, which is where forensic IT review of downloads, email forwarding and device activity earns its keep; and speed. Delay is close to fatal — a court asked for urgent relief will want to know why the employer waited six weeks.

For the employee

The defences track the doctrine. No legitimate interest: the clause protects against mere competition. Overbreadth: too long, too wide, or covering activities the employee never performed — tested at the date of signing. Personal connections: the clients followed the individual, not the firm, and some were the employee's before they arrived. Stale or public information: what the employer calls confidential is industry knowledge or the employee's own skill. And a defence with a long pedigree: if the employer itself repudiated the contract — dismissing wrongfully, or committing a serious breach the employee accepted as ending the employment — the employer generally cannot hold the employee to the restraint. A departure that also raises a dismissal grievance is a different fight, covered in our unfair dismissal guide.

The announced federal ban — not yet law

In the 2025-26 Federal Budget, the Commonwealth Government announced its intention to ban non-compete clauses for most employees earning below the high-income threshold in the Fair Work Act 2009 (Cth) — a figure indexed each year. The change is planned to commence from 2027 and to operate prospectively, with consultation continuing on details such as penalties, the treatment of non-solicitation and non-poaching clauses, and whether restraints on higher earners should also be constrained.

Two cautions. First, this is announced reform, not current law: as matters stand, the common law described above — and in New South Wales the 1976 Act — still governs, and restraints signed today are still tested on ordinary principles. Second, the announcement targets non-competes for workers under the threshold; confidentiality obligations and non-solicitation covenants are, for now, a separate question. Anyone signing, drafting or litigating a restraint in 2026 should proceed under existing law while watching the legislation.

Practical notes for both sides

If you are the employer: draft narrow and defensible, rather than wide and hopeful. Anchor each covenant to a named interest; prefer non-solicitation and non-dealing over blanket non-competes; use a short, sane cascade rather than a combinatorial monster; refresh restraints on every promotion, supported by fresh consideration; choose New South Wales law where there is a genuine connection; and protect confidential information with systems — access controls, exit audits, prompt device recovery — because a clause is not a firewall.

If you are the employee: negotiate the restraint before you sign, when you have leverage, not after you resign, when you have none. On exit, ask what the employer actually intends to enforce, and seek a written release or a narrowed restraint in the departure deed — employers trade restraint scope for a clean handover more often than people expect. Take nothing with you: a downloaded client list converts a shaky restraint case into a strong confidence case. Think hard before giving undertakings, which bind like an injunction — and negotiate on a without prejudice basis so offers cannot be used against you.

Restraint cases reward whichever side understands the doctrine first. The clause on the page is the beginning of the analysis, not the end: what a court will actually enforce depends on the interest, the role, the drafting, the state — and increasingly, on legislation still taking shape in Canberra.

This article is published by CommonBench for informational purposes only and does not constitute legal advice. If you are deciding whether the restraint in an Australian employment contract would actually hold up — or how far you can safely enforce one — try CommonBench — AI-powered legal research with verified citations across five common law jurisdictions.