Continuation Claims: What the Federal Circuit Has Held

Continuation practice runs on a single tension. An applicant may pursue claim boundaries nobody selected at filing, informed by products and markets that emerged later. The claimed subject matter must still have been disclosed in the specification as originally filed, and inventorship still turns on human conception of what is claimed. Most of the reported failures come from drafters who relied on the first proposition and did not test against the second.

What follows is a reading of the decided cases, organized by the question each one answers. It is a starting point for a practitioner's own analysis, not a substitute for it.

Last updated: July 2026. Educational only, not legal advice, and not a prediction for any specific application. Primarily US law and USPTO practice; other jurisdictions differ. Case law changes; confirm current treatment before relying on anything here.

Related practitioner references

What the case law permits

Claims may be drafted to read on a competitor's known product

In Kingsdown Medical Consultants v. Hollister, 863 F.2d 867 (Fed. Cir. 1988), the Federal Circuit addressed the practice directly: filing an application to obtain the right to exclude a known competitor's product is not, by itself, improper or inequitable conduct. State Industries v. A.O. Smith is cited for the related proposition that keeping an application alive to cover a competitor's product is legitimate.

This is worth stating plainly because clients often assume otherwise. Under these holdings the constraint on such a claim is not the drafter's intent. It is whether the specification supports the claim.

Disclosed but unclaimed subject matter can be dedicated to the public

This is the affirmative case for continuation practice, and it comes from Johnson & Johnston Associates v. R.E. Service, 285 F.3d 1046 (Fed. Cir. 2002) (en banc). The specification named aluminum, stainless steel, nickel alloy, and polypropylene substrates. The claims recited aluminum. The accused product used steel. The court held that "when a patent drafter discloses but declines to claim subject matter, as in this case, this action dedicates that unclaimed subject matter to the public," foreclosing recovery under the doctrine of equivalents.

The court identified the remedies: a reissue application under 35 U.S.C. §251 within two years, or a separate application claiming the disclosed subject matter under §120, filed before every application in the chain issues. The patentee in that case had in fact filed continuations claiming the steel substrate.

The doctrine is narrower than it is often paraphrased, and the precision matters. Dedication principally forecloses recapturing a disclosed alternative through the doctrine of equivalents. It does not place the subject matter beyond reach while the family is still pending: a continuation can still seek literal claims to that alternative, subject to written description, enablement, prior art, eligibility, double patenting, and remaining term. What closes is the opportunity, not merely a theory of infringement.

The implication for a continuation claim set is structural: the alternatives, materials, ranges, and embodiments a specification names but the issued claims do not reach are the subject matter whose literal coverage becomes unavailable once the last application in the family issues or goes abandoned.

Where continuation claims have failed

Six recurring failure modes, each anchored to a holding.

1. Written description: the specification must describe the claimed invention, not merely make it obvious

In ICU Medical v. Alaris Medical Systems, 558 F.3d 1368 (Fed. Cir. 2009), claims to a "spikeless" medical valve were added years after filing with the specification unchanged. Every disclosed embodiment had a spike. The court held the claims invalid for lack of written description, and rejected the argument that a skilled artisan's understanding could bridge the gap: "It is not enough that it would have been obvious to a person of ordinary skill that a preslit trampoline seal could be used without a spike." An applicant complies "by describing the invention, with all its claimed limitations, not that which makes it obvious."

Gentry Gallery v. Berkline, 134 F.3d 1473 (Fed. Cir. 1998), is the same failure in a different technology. The disclosure put the recliner controls on a fixed console; claims that removed that location were invalid. Claims may be no broader than the supporting disclosure, so a narrow disclosure limits claim breadth.

The common shape: a continuation claim that omits a limitation present in every disclosed embodiment.

The converse error is also worth naming: support is not a search for identical words. The statutory test is whether the disclosure reasonably conveys that the inventor had possession of the claimed subject matter as of the filing date, and MPEP 2163 directs that the disclosure be assessed as a whole, including drawings and the relationships they show. A claim can therefore be supported without reciting the specification's exact terminology. The inverse also holds, and is the more dangerous case for automated review: every individual term of a claim can appear somewhere in a specification while the claimed combination appears nowhere, because the elements were drawn from embodiments the disclosure never joined. Word-level matching produces both false negatives and false positives on this question.

2. Enablement must reach the full scope of the claim

In Liebel-Flarsheim v. Medrad, 481 F.3d 1371 (Fed. Cir. 2007), the claims were construed to cover injectors with or without a pressure jacket. The specification taught away from the jacketless version, and the evidence showed it could not have been built at the time of filing. The claims were invalid for lack of enablement.

The case is cited for the proposition that a disclosure must teach the full range of embodiments a claim covers. A broad construction won in litigation can supply the challenger with an enablement defense.

3. A disclaimer made in the parent carries into the continuation

Under Hakim v. Cannon Avent Group, 479 F.3d 1313, 1318 (Fed. Cir. 2007), an applicant may file a continuation with broader claims, but a disclaimer made during the parent's prosecution continues to bind unless the prosecution history is "sufficiently clear to inform the examiner that the previous disclaimer, and the prior art that it was made to avoid, may need to be re-visited." Traxcell Technologies v. Nokia, 15 F.4th 1136, 1146 (Fed. Cir. 2021), restates that claim scope is disclaimed by argument or by amendment.

Practitioner commentary describes an express statement on the record rescinding prior disclaimers, sometimes called a Hakim statement, as the response to this. What qualifies as "sufficiently clear" has not been closely delineated by the courts, so the effect of any particular statement remains a litigated question.

The effect also runs in the other direction, and that is the more recent development. In Barrette Outdoor Living v. Fortress Iron, 156 F.4th 1353 (Fed. Cir. Oct. 17, 2025), statements made while prosecuting a later-filed continuation limited the scope of patents in the same family that had already issued. The court built on Teva Pharmaceuticals USA v. Sandoz, 789 F.3d 1335 (Fed. Cir. 2015), under which prosecution statements bear on claim construction regardless of whether they predate or postdate issuance of the patent being construed.

Taken together, Hakim and Barrette describe a family record that is continuous in both directions. A characterization offered to secure one continuation can travel backward to a sibling that issued years earlier.

4. Functional claim terms may be read as means-plus-function

Continuation claims often reach for functional language to cover an implementation that differs from the disclosed one. Williamson v. Citrix Online, 792 F.3d 1339 (Fed. Cir. 2015), sitting en banc on this issue, "expressly overrule[d] the characterization of that presumption as 'strong'" for terms that do not use the word "means." It held "distributed learning control module" was a means-plus-function term lacking corresponding structure, and therefore indefinite, observing that "module" is a well-known nonce word that operates as a substitute for "means."

Under §112(f), such a term is limited to the structure disclosed in the specification and its equivalents, which is often narrower than the functional language suggests, or indefinite where no structure is disclosed.

5. The §121 safe harbor turns on the application's relationship to the restriction, not on a label applied later

The safe harbor in 35 U.S.C. §121 protects applications filed as a result of a restriction requirement, and the Federal Circuit has limited that protection to divisionals. Pfizer v. Teva, 518 F.3d 1353 (Fed. Cir. 2008), denied it to a continuation-in-part, reasoning that a CIP can contain matter beyond the originally restricted claims. G.D. Searle v. Lupin Pharmaceuticals (Fed. Cir. 2015) held that an application prosecuted as a CIP rather than a divisional could not be repaired by reissue.

The protection depends on the application's substantive relationship to the restriction requirement, on its filing history, and on continued consonance with the restricted invention groups. Relabeling an application as a divisional later does not by itself create the protection. This is a filing posture rather than a claim-drafting choice, but it governs the double-patenting exposure of every claim in the resulting application.

6. Prosecution delay can render claims unenforceable

Hyatt v. Hirshfeld, 998 F.3d 1347 (Fed. Cir. 2021), states a two-part test for prosecution laches: delay that is unreasonable and inexcusable under the totality of the circumstances, plus prejudice attributable to that delay. Personalized Media Communications v. Apple applied it to hold a patent unenforceable, in a chain running from a 1981 parent through a 1987 continuation-in-part to a 1995 application.

The 2025 decisions emphasize the second element. In Google v. Sonos, No. 2024-1097 (Fed. Cir. Aug. 28, 2025), the court reversed a district court's laches ruling and reinstated the verdict, stating that standard continuation practice, even sustained over more than a decade, is not itself an egregious misuse of the system where the accused infringer cannot show prejudice. The claims there had not been presented in claim form until 2019. That decision is nonprecedential, so it is persuasive rather than binding, and it should be read as such.

The synthesis the cases support is not that long continuation chains are inherently improper. It is that prosecution laches is an exceptional equitable doctrine requiring both unreasonable, unexplained delay and demonstrated prejudice. A separate precedential decision, Hyatt v. Stewart, No. 18-2390 (Fed. Cir. Aug. 29, 2025), issued a day later in the long-running Hyatt litigation; this page does not characterize its holding, which should be read directly.

A note on recapture

The recapture rule is a reissue doctrine. It bars a broadening reissue from retrieving scope surrendered during prosecution, under the three-step analysis of In re Clement and MPEP 1412.02. A continuation filed before issuance pursues broader claims without reaching that doctrine. The nuance worth carrying: for a later reissue, surrender is assessed across the family, including scope given up in a preceding divisional, continuation, or continuation-in-part. Continuations avoid the recapture rule while still contributing to the surrender record. The separate constraint on a continuation is the Hakim disclaimer analysis above, and, for equivalents, prosecution history estoppel under Festo, 535 U.S. 722 (2002), which can arise from an amendment made to satisfy any requirement of the Patent Act, not only a prior-art amendment.

Kroy and the value of a materially different claim

In Kroy IP Holdings v. Groupon, No. 2023-1359 (Fed. Cir. Feb. 10, 2025), the court held that a PTAB determination that some claims of a patent are unpatentable does not collaterally estop the patentee from asserting claims the IPR never adjudicated. The reasoning turns on the standard of proof: unpatentability is shown at the PTAB by a preponderance of the evidence, while invalidity in district court requires clear and convincing evidence, and a party that carried the lower burden is not entitled to assert preclusion where the higher one applies.

The scope of that holding was tested and confirmed. A petition for panel rehearing and rehearing en banc, which asked whether the panel had erred in permitting assertion of claims that do not differ materially from those held unpatentable, was denied on August 1, 2025, per curiam, with Chief Judge Moore concurring and Judge Dyk dissenting. The dissent's objection was that the rule permits patentees to evade IPR outcomes through marginally different claims.

Two limits are worth noting alongside it. The rationale is the difference in burden, not the difference between claims, so it does not extend to a prior district court invalidity judgment, where the standard is the same in both proceedings and Ohio Willow Wood v. Alps South continues to govern patentably indistinct claims. And preclusion is a procedural bar, not a merits defense: claims that survive estoppel can still be invalidated on the same art.

Read together with the double-patenting analysis on the continuations page, the decided cases point in a consistent direction. A continuation claim set that differs from the parent by word choice, by reordering steps, or by moving a limitation between the preamble and the body tends to share the parent's fate on the merits. Claims directed to a different combination, a different actor, a different point in the system, or a different statutory class are the ones assessed on their own limitations. Material difference does not make a claim valid. It means the claim can stand or fall on its own rather than being resolved by what happened to its siblings.

Sources