Claim Construction

Claim construction is the judge’s determination of what the words in a patent’s claims legally mean. It matters because infringement and invalidity are both measured against the claims, not against the patent’s drawings, its abstract, or the product the inventor actually sold. Fix the meaning of a handful of words and you have usually fixed the outcome of the lawsuit.

The proceeding where this happens is called a Markman hearing, after Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996), in which the Supreme Court held that construing a patent claim is a question for the court rather than the jury. In most district courts the Markman order lands months before trial, and in a large share of cases nothing meaningful happens afterward except settlement.

Why claim words need construing at all

A patent’s claims are numbered sentences at the end of the document that mark the boundary of the monopoly, the way a deed marks a parcel. Infringement under 35 U.S.C. § 271 requires that the accused product or process contain every element of at least one claim. Invalidity under §§ 102 and 103 asks whether the prior art contained those same elements first. Neither question is answerable until you know what the elements are. If a claim requires a “pipe,” does that cover a rectangular duct? If it requires a component “mounted on” a board, does one soldered through the board qualify? Claim drafters write in deliberately general language to capture future variations, and the generality that gives a patent its reach is the same generality that makes its edges arguable.

Claims are construed from the perspective of a person of ordinary skill in the art, as of the patent’s effective filing date. That is a legal construct, not the accused infringer and not the judge. A term of art therefore carries its technical meaning in the field even when ordinary English suggests something else, and later industry usage does not retroactively change what a claim covers.

Who decides, and why the jury does not

Before 1996 courts split on whether claim meaning was a fact question the Seventh Amendment reserved for juries. Markman resolved it. Justice Souter’s opinion for a unanimous Court treated the patent as a legal instrument closer to a deed than to a disputed contract, found no clear 18th-century practice of jury construction, and concluded that judges are better positioned to read a technical document and that judicial construction promotes uniformity. The same words in the same patent should mean the same thing in the Eastern District of Texas and the Northern District of California.

The division of labor that follows is clean in theory. The judge says what the claim means. The jury decides whether the accused product meets the claim as construed, and whether the prior art did. Under O2 Micro International Ltd. v. Beyond Innovation Technology Co., 521 F.3d 1351 (Fed. Cir. 2008), a court cannot dodge a genuine dispute about claim scope by handing the jury the raw claim language and telling it to apply the “plain and ordinary meaning.” Resolving the dispute is the court’s job.

The Phillips hierarchy of evidence

The controlling methodology comes from Phillips v. AWH Corp., 415 F.3d 1303 (Fed. Cir. 2005) (en banc), which rejected a dictionary-first approach that had crept into the case law and restored the patent document to primacy. The ordering is:

• The claims themselves, including the other claims. Claim differentiation presumes an independent claim is broader than a dependent claim that adds a limitation, so a term cannot usually be read to already include what a dependent claim separately requires.
• The specification. Phillips called it “the single best guide to the meaning of a disputed term,” while warning against the opposite error of importing a limitation from an embodiment into a claim that does not recite it. That tension is where most Markman disputes actually live.
• The prosecution history, meaning what the applicant told the USPTO to get the patent allowed. Arguments and amendments made to distinguish prior art can surrender claim scope, and the patentee does not get it back in litigation.
• Extrinsic evidence: dictionaries, treatises, expert and inventor testimony. Admissible and sometimes useful for technical background, but Phillips held it is “less significant than the intrinsic record” and cannot contradict a meaning the patent itself makes clear.

Two escape hatches move a term off its ordinary meaning, and Thorner v. Sony Computer Entertainment America LLC, 669 F.3d 1362 (Fed. Cir. 2012), sets the bar high for both: lexicography, where the patentee acts as their own dictionary and clearly defines a term, and disavowal, where the patentee clearly and unmistakably disclaims scope. Vague hints do not count.

What actually happens at a Markman hearing

Local patent rules vary a great deal, but the shape is consistent. The parties exchange proposed constructions for the terms they dispute, narrow the list, and file a joint claim construction statement identifying what remains. Briefing follows, usually with intrinsic-record appendices and sometimes competing expert declarations on how a skilled artisan would read the term. The hearing itself may run an hour or a full day, often with a technology tutorial for the judge.

The order can do more than assign meanings. It can hold a term indefinite under 35 U.S.C. § 112(b), which kills the claim outright; Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898 (2014), asks whether the claim informs a skilled person of the invention’s scope with reasonable certainty. It can hold a term to be a means-plus-function limitation under § 112(f), which under Williamson v. Citrix Online, LLC, 792 F.3d 1339 (Fed. Cir. 2015), confines the claim to structure actually disclosed in the specification and its equivalents, and invalidates the claim if no such structure was disclosed. Functional software claims fail this way with some regularity.

The Patent Trial and Appeal Board also construes claims in inter partes review, and since a 2018 rule change it applies the same Phillips standard the district courts use, rather than the broader “broadest reasonable interpretation” it once used.

Why the hearing so often ends the case

Claim construction is a vise. The construction broad enough to capture the accused product tends to be broad enough to read on the prior art, which invalidates the claim. The construction narrow enough to clear the prior art tends to exclude the accused product, which defeats infringement. Both sides know this going in, which is why the disputed terms are chosen with the summary judgment motion already drafted in mind.

So everything before the Markman order is preparation for it, and much of what follows is bookkeeping. A losing patentee will frequently stipulate to non-infringement in order to take an immediate appeal on the construction. A losing defendant will frequently settle. Cases that reach a jury on the merits are the minority.

Appeals and the Teva standard of review

For years the Federal Circuit reviewed claim construction entirely de novo, giving the trial judge no deference at all, under Cybor Corp. v. FAS Technologies, Inc., 138 F.3d 1448 (Fed. Cir. 1998) (en banc). Reversal rates were high enough that no construction felt final until the appeal was over.

Teva Pharmaceuticals USA, Inc. v. Sandoz, Inc., 574 U.S. 318 (2015), changed part of that. The Supreme Court held that Federal Rule of Civil Procedure 52(a)(6) applies to patent cases like any other, so subsidiary factual findings a district judge makes when weighing extrinsic evidence are reviewed only for clear error. The ultimate construction remains a legal conclusion reviewed de novo.

The gap this opened matters strategically. Where a construction rests only on the intrinsic record, the Federal Circuit still starts from scratch. Where it rests on findings about disputed extrinsic facts, those findings are much harder to dislodge. That gives a party who wins below a reason to build a factual record, and gives the party who expects to lose a reason to keep the dispute purely intrinsic.

Frequently asked questions

What is claim construction in a patent case? Claim construction is the process of determining the legal meaning and scope of the words in a patent’s claims. Because infringement means the accused product contains every element of a claim, and invalidity means the prior art does, the meaning of the claim words controls both questions. Under Markman v. Westview Instruments (1996), a judge decides claim construction as a matter of law, not a jury.

What is a Markman hearing? A Markman hearing is the pretrial proceeding where a district judge hears argument, and sometimes expert testimony, on the disputed claim terms and then issues an order defining what those terms mean. It takes its name from Markman v. Westview Instruments. It is usually held after fact discovery on the patent and before summary judgment, and the resulting order binds the rest of the case.

What evidence does a court use to construe patent claims? Phillips v. AWH Corp. (Fed. Cir. 2005, en banc) sets a hierarchy. Intrinsic evidence comes first: the claim language itself, then the specification, then the prosecution history. Extrinsic evidence such as dictionaries, treatises, and expert testimony may be consulted but cannot override the intrinsic record. The touchstone is how a person of ordinary skill in the art would have read the term at the time of filing.

Why do Markman hearings often decide the case? A construction that is broad enough to cover the accused product often also reads on the prior art and invalidates the claim; a construction narrow enough to survive the prior art often excludes the accused product. Once the judge fixes the meaning, one side frequently concedes or loses on summary judgment, so many cases settle within weeks of the Markman order.

Authorities and sources

• Markman v. Westview Instruments, Inc., 517 U.S. 370 (1996) (Cornell LII)
• Teva Pharmaceuticals USA, Inc. v. Sandoz, Inc., 574 U.S. 318 (2015) (Cornell LII)
• Nautilus, Inc. v. Biosig Instruments, Inc., 572 U.S. 898 (2014) (Cornell LII)
• 35 U.S.C. § 112: Specification, including definiteness and means-plus-function (Cornell LII)
• 35 U.S.C. § 271: Infringement of patent (Cornell LII)
• 35 U.S.C. § 102: Conditions for patentability, novelty (Cornell LII)
• 35 U.S.C. § 103: Conditions for patentability, non-obvious subject matter (Cornell LII)
• Federal Rule of Civil Procedure 52: Findings and conclusions by the court (Cornell LII)
• MPEP 2111: Claim interpretation, broadest reasonable interpretation (USPTO)
• USPTO final rule: claim construction standard in PTAB trials (83 Fed. Reg. 51340, Oct. 11, 2018) (GovInfo)

Going further: What patent litigation actually costs, and how long it takes .

This page is general legal information, not legal advice, and it does not create an attorney-client relationship.