New York has no law banning non-competes, which means there is no single answer and no clause that is automatically void. A New York court looks at the particular agreement, the particular job, and what the employer is actually trying to protect, and the standard it applies is narrow enough that a lot of broadly written clauses do not survive it.
Checked 23 October 2026. This area has moved quickly. The federal ban that was widely reported in 2024 never took effect and has since been removed from the regulations, and a New York bill passed the State Senate in June 2026 but has not become law. Both are covered further down, because people usually arrive here having heard about one of them.
The test a New York court applies
The governing case is BDO Seidman v. Hirshberg, decided by the New York Court of Appeals in 1999 and still the standard. The court set out three conditions, and its own wording is worth reading rather than paraphrasing. A restraint is reasonable only if it:
- →"is no greater than is required for the protection of the legitimate interest of the employer"
- →"does not impose undue hardship on the employee"
- →"is not injurious to the public"
The sentence that follows is the one that matters most to anyone holding a non-compete: "A violation of any prong renders the covenant invalid." An employer has to clear all three, and the first one is where most agreements run into trouble.
What counts as a legitimate interest
New York draws this narrowly. The Court of Appeals in BDO Seidman describes limiting "the cognizable employer interests under the first prong of the common-law rule to the protection against misappropriation of the employer’s trade secrets or of confidential customer lists, or protection from competition by a former employee whose services are unique or extraordinary".
Three things, then. Trade secrets. Confidential customer lists. Or an employee whose services were genuinely unique or extraordinary. Not wanting a competitor to benefit from your training, not disliking the idea of you working down the road, not general protection against competition — none of those is on the list, and a clause that protects nothing beyond them has a problem at the first prong.
The unique-or-extraordinary route is harder to use than employers tend to assume. Being good at your job does not get there. Courts have applied it to people whose particular talent or relationships were themselves the thing clients were buying, which is why it appears more often in cases about professionals, performers and senior figures than about ordinary staff. A customer list argument also tends to collapse where the contacts are public knowledge or the client was yours before you arrived.
Why an overbroad clause is not simply dead
A common assumption is that a clause stretched too far becomes unenforceable in its entirety. New York courts can take a different route. BDO Seidman confirms "the judicial power to sever and grant partial enforcement for an overbroad employee restrictive covenant", and the Court of Appeals criticised the lower court in that case precisely for refusing to do it.
So a three-year nationwide restriction can come out of court as a one-year restriction covering the clients you personally served. That is a much better outcome than full enforcement and a much worse one than nothing, and it is why the question is rarely as binary as either side first presents it.
The federal ban people remember
In April 2024 the Federal Trade Commission issued a rule that would have voided most non-competes across the country, with an effective date in September that year. A federal court in Texas set it aside before it took effect. The Commission appealed and then withdrew those appeals in September 2025, and the rule was formally removed from the Code of Federal Regulations on 12 February 2026.
It never applied to anybody. If you signed a non-compete in 2024 believing it had been cancelled by Washington, it was not. The FTC has continued to act against individual employers under its general authority, which is a slower and much narrower kind of pressure, and it does nothing for your agreement.
The New York bill
Albany has come close twice. A broad ban passed the legislature in 2023 and Governor Hochul vetoed it that December, objecting among other things to the absence of any salary threshold and to its reach over agreements already signed. A narrower version with a $500,000 earnings cut-off and prospective effect only passed the Senate in 2025 without reaching an Assembly vote, and a substantively similar bill passed the Senate again on 3 June 2026.
As at the date on this page it sits in the Assembly Labor Committee and has not been enacted. Treat it as something to watch rather than something to rely on, and note that the versions put forward since the veto have applied only going forward — a ban written that way would not reach a non-compete you have already signed.
Reading the one you have
Start by working out what the clause actually restricts, because the three documents people lump together do different jobs and the broad one is usually the weakest.
| Clause | What it stops you doing | How it tends to fare |
|---|---|---|
| Non-compete | Working for a competitor, or in the industry, for a period and within an area | The hardest to enforce in New York, because it has to clear all three prongs and usually protects nothing on the narrow list of legitimate interests |
| Non-solicit of clients | Approaching customers you dealt with | Much stronger ground, particularly where it is limited to clients you personally served rather than the employer's whole book |
| Non-solicit of staff | Recruiting former colleagues | Commonly enforced, and rarely the clause worth fighting about |
| Confidentiality | Using or disclosing the employer's information | Enforceable and usually uncontroversial. It survives separately, so losing a non-compete does not free you to take files |
Then check the duration, the geography and the definition of the restricted business. A clause covering an entire industry nationwide for two years invites the argument that it goes further than any legitimate interest requires. One limited to the accounts you handled, for a year, within the area you worked, is the version an employer is more likely to get enforced.
If a specific sentence is doing the damage, our contract clause analyzer will take that clause on its own and explain in plain English what it does and who it favours. It cannot tell you how a New York judge would rule on your facts, which is the question that actually decides it.
What usually happens in practice
Litigation is expensive and slow, and most employers never file. The clause does its work by deterring the employer who might hire you, which is why people discover theirs matters at the point a new company’s HR department asks whether they are subject to one.
That reframes what is worth doing. A written release, or a letter from your former employer confirming it will not enforce, solves the hiring problem without any court involvement, and companies grant them more readily than you might expect when the departure was amicable. Asking for one on your way out costs nothing. Asking after you have a competing offer is harder.
Where a job is genuinely at stake, an hour with a New York employment lawyer is worth the money. They will have seen how local courts treat clauses in your industry, which is the part no general guide can supply, and they can tell you quickly whether yours is the kind that gets partially enforced or the kind that falls at the first prong.
The short version for New York: there is no ban, the federal rule never arrived, and the test is three conditions an employer has to meet in full. Most arguments are won or lost on the first of them, which asks what the employer is really protecting — and for most ordinary jobs, the honest answer falls outside the narrow list the Court of Appeals allows.
General legal information, not legal advice. This describes New York law as at 23 October 2026 and nothing else; other states take very different approaches, and California voids nearly all employee non-competes outright. Outcomes turn on facts a page like this cannot see. BDO Seidman v. Hirshberg, 93 N.Y.2d 382 (1999), is quoted from the opinion and can be read in full on CourtListener.
Free, two analyses a day, no account needed.
Editorial note: AI For Legal Research publishes independent content. We do not accept payment for editorial coverage or review scores. Nothing on this site constitutes legal advice. Always consult a qualified attorney for legal matters.