How long does a patent last? For a U.S. utility patent, the answer is set by 35 U.S.C. 154: the term runs from the date the patent issues until 20 years after the application was filed in the United States. The detail that trips people up is the measuring point. The 20 years is counted from the filing date, not the issue date, so a patent that took four years to prosecute has roughly sixteen years of enforceable life left when it issues, while one that issued quickly has closer to the full twenty. The grant date opens the enforcement window; the filing date closes it twenty years later.

The statute is more specific still when a patent is part of a family. If the application contains a specific reference to an earlier-filed U.S. application whose benefit it claims, the 20 years is measured from that earliest such filing date — not from the later child's own filing. Section 154(a)(2) states the term as:

ending 20 years from the date on which the application for the patent was filed in the United States or, if the application contains a specific reference to an earlier filed application or applications under section 120, 121, 365(c), or 386(c), from the date on which the earliest such application was filed.— 35 U.S.C. § 154(a)(2), source

That clause is why a continuation or divisional filed years after the original does not get a fresh 20-year clock. Sections 120, 121, 365(c), and 386(c) are the benefit-claiming provisions — continuations, divisionals, continuations-in-part, and national-stage and design counterparts. When a patent claims benefit under any of them, its term is pinned to "the earliest such application" in the chain. A family of ten patents all tracing back to one 2010 parent will, as to that benefit chain, all run out around 2030, regardless of how late the individual children were filed. For an IP reader mapping a competitor's portfolio, this is the difference between a thicket that expires together and one staggered across many years.

What can move the effective expiration date

The 20-years-from-filing figure is the statutory baseline, but the effective expiration of a particular patent can sit earlier or later for reasons the same statutory scheme provides. Patent term adjustment, also governed by section 154, can extend the term to compensate for certain delays attributable to the USPTO during prosecution — the patent gets back some of the time the office took beyond defined deadlines. Moving the other way, a terminal disclaimer (filed to overcome an obviousness-type double-patenting rejection) can cut a patent's term so that it expires no later than a related patent, and any maintenance-fee lapse can end it early. A separate provision, 35 U.S.C. 156, allows term extension for regulatory review delay in specific contexts, which is a different mechanism from the 154 adjustment. The point for a reader is that the printed-on-the-face calculation starts at 20 years from the earliest claimed filing, and then the adjustments and disclaimers, if any, modify it.

Why filing-date measurement changed the incentives

The 20-years-from-filing rule is a relatively modern feature of U.S. law; for patents filed before mid-1995 the term ran from the grant date instead, on a different number of years. Measuring from filing rather than from grant removed an incentive that the older grant-date system created — the incentive to prolong prosecution, because under a grant-date term, delay at the office pushed the expiration later and lengthened the monopoly. Under the current section 154 rule, delay does the opposite: every additional year in prosecution is a year subtracted from enforceable life, because the 20-year endpoint is fixed at the filing date and does not move when the grant slips. That is precisely why Congress paired the filing-date term with the patent term adjustment mechanism, restoring time lost specifically to USPTO-caused delay so that applicants are not penalized for the office's own backlog. The reader's takeaway is that the modern term structure is built to make the patentee neutral to office delay — neither rewarded nor punished by it — while keeping the outer boundary pinned to the earliest claimed filing date. Reading any post-1995 utility patent therefore means starting from filing, not grant, and treating term adjustment as the correction for office-side delay rather than as a bonus.

Reading the term off a real record

The mechanics are visible on any granted record. Take United States Patent US10804246B2, "Microelectronics package with vertically stacked dies" — a real granted semiconductor packaging patent. To compute its baseline term you do not look at the grant date; you find the earliest U.S. application it claims benefit to under section 120 (its priority chain) and add 20 years to that date, then apply any term adjustment shown on the record and check for a terminal disclaimer. The same procedure applies to every utility patent in a portfolio: identify the earliest claimed U.S. filing, add twenty years, adjust. That is the calculation the statute prescribes, and it is the one a competitive-intelligence read should run before treating any patent as a long-lived obstacle.

What the record shows is the statutory rule and its measuring point. 35 U.S.C. 154 grants a U.S. utility patent a term beginning at issuance and ending 20 years from the date the application was filed in the United States, or from the earliest filing date of an application whose benefit it specifically claims under sections 120, 121, 365(c), or 386(c). The clock runs from filing, not from grant; priority claims pull it back to the earliest application in the chain; and term adjustment for office delay, terminal disclaimers, and maintenance-fee status can shift the effective expiration. The statute does not promise any particular number of enforceable years after grant — that depends entirely on the filing date and the family. For a reader assessing how long a competitor's claim will stand, the first move is always to find the earliest claimed U.S. filing date and count forward twenty years.