Prior art is any evidence that an invention was already known before a patent application's effective filing date. It is the single most important concept in patent validity, and the one most often misunderstood by people encountering it for the first time.
The test is public availability, not awareness. A thesis sitting in a German university library, never cited, read by nobody, is prior art. So is a product manual, a conference poster, and a dated forum post.
What qualifies under 35 U.S.C. 102
| Category | Examples | Notes |
|---|---|---|
| Patents | Issued patents, any country | Full disclosure counts, not just claims |
| Published applications | Any office, any language | Published at 18 months by default |
| Printed publications | Journals, conference papers, theses, manuals | Must be publicly accessible |
| Public use | Demonstrations, field trials, deployed products | In the US or anywhere |
| On sale | Sales and offers for sale | Includes the inventor's own |
| Otherwise available | Websites, videos, standards documents | Catch-all added by the AIA |
Geography and language do not limit it. Under the current statute, a Chinese utility model or a Japanese conference paper is prior art against a US application exactly as an American patent would be. This was not always true — pre-AIA law contained geographic limits — and applications with early priority dates are still assessed under the old rules.
The whole disclosure counts, not just the claims. A patent that claims one thing but describes another in its specification is prior art for everything it describes. This is why the closest prior art is frequently found in the body of a patent whose claims are about something else entirely.
Public accessibility is the threshold. A document is a printed publication if an interested person exercising reasonable diligence could have found it. Cataloguing in a library satisfies this. Circulation among a closed group under confidentiality generally does not.
Anticipation versus obviousness
Prior art defeats a claim two ways, and the difference matters enormously in practice.
| Anticipation (§102) | Obviousness (§103) | |
|---|---|---|
| References needed | One | Several may be combined |
| Requirement | Every element, arranged as claimed | Reason to combine, reasonable expectation of success |
| Also requires | Enablement of the reference | Consideration of secondary factors |
| Difficulty to establish | Higher | Lower |
| Difficulty to rebut | Higher | Lower |
Anticipation is strict. A single reference must disclose every element of the claim, arranged as the claim arranges them. A reference that discloses four of five elements does not anticipate, however obvious the fifth might be.
Obviousness allows combination, subject to a reason to combine. After KSR v. Teleflex (2007), that reason can come from market forces, design incentives or common sense rather than an explicit teaching in the references — which made obviousness rejections considerably easier to make and harder to overcome.
Secondary considerations can rebut obviousness. Commercial success, long-felt need, failure of others, unexpected results and copying are all relevant, and a nexus between the evidence and the claimed features must be shown.
Your own disclosure
This catches inventors more than any other rule in patent law.
The US grace period. Under 35 U.S.C. 102(b)(1), an inventor's own public disclosure does not count as prior art against their own application if the application is filed within twelve months of that disclosure.
Most of the world has no grace period. A public disclosure before filing destroys novelty immediately in Europe, China and most other jurisdictions. An inventor who presents at a conference and files a US application eight months later has preserved US rights and lost European ones.
The safe sequence is file first, disclose second, which is why patent prosecution usually begins with a provisional. A provisional application costs relatively little and establishes a priority date before any conference paper, demonstration, sales pitch or funding presentation.
What counts as disclosure is broader than people expect:
| Activity | Prior art risk |
|---|---|
| Conference presentation | Yes |
| Journal article or preprint | Yes |
| Public demonstration or trade show | Yes |
| Offer for sale, even without disclosing details | Yes — on-sale bar |
| Kickstarter or crowdfunding campaign | Yes |
| Public GitHub repository | Yes |
| Pitch under NDA | Generally no |
| Internal use within a company | Generally no |
The on-sale bar deserves particular attention. A commercial offer for sale triggers it even where the invention itself was never disclosed in the transaction. A quote sent to a customer describing a product not yet built can start the clock.
What a prior art search covers
Patent literature across the major offices — USPTO, EPO, WIPO, JPO, CNIPA and KIPO — searched by classification and by keyword. Classification search matters because terminology varies wildly between fields and eras, and a keyword search alone misses references that describe the same thing in different words.
Non-patent literature, which is where the damaging references usually hide. Journals, conference proceedings, standards documents, technical manuals, theses, and product datasheets. This category is harder to search systematically and is the one most often skipped in a cursory search.
The date logic is unforgiving. Everything published before the effective filing date is potentially relevant, and anything published after it is not. Establishing publication dates for non-patent literature — particularly web content — is frequently the hardest part of the exercise.
Search types and what each is for
| Search | Question | Date scope | Expired patents |
|---|---|---|---|
| Patentability / novelty | Can this be patented? | Everything before filing | Relevant |
| Validity | Can this patent be invalidated? | Everything before priority | Relevant |
| Freedom to operate | Can we sell this? | In-force claims only | Irrelevant |
| State of the art | What exists in this field? | Everything | Relevant |
These are not interchangeable. The most common and most costly confusion is treating a patentability search as freedom-to-operate clearance. The first asks whether an invention is new; the second asks whether a product infringes. An expired patent is fatal to the first question and completely irrelevant to the second.
A worked example: what a validity search produces
A patent claims a method with five elements: A, B, C, D and E. The search produces four references.
| Reference | Discloses | Date | Effect |
|---|---|---|---|
| US patent | A, B, C, D | 4 years before priority | Does not anticipate — E missing |
| Journal article | E, in the same field | 2 years before priority | Combinable with the patent |
| Conference paper | A, B, C, D, E | 8 months after priority | Not prior art |
| Product manual | A, B, C, D, E | 6 years before priority | Anticipates |
The product manual is the finding. A single reference disclosing every element, published before the priority date, anticipates under 102 — and it is non-patent literature, which is exactly the category an examiner is least likely to have searched.
The patent and article together support an obviousness argument, but that is weaker: it requires establishing a reason to combine, and the patentee can respond with secondary considerations.
The conference paper is worthless despite being a perfect match, because it postdates the priority date. Date discipline is the whole exercise.
Note what makes the manual valuable in a PTAB proceeding. A reference the examiner never considered raises a substantial new question of patentability. One already of record does not, which is why searches for invalidation deliberately target what the file history shows was missed.
Prior art in an IPR
Inter partes review is limited to patents and printed publications under 35 U.S.C. 311(b). Public use, on-sale activity and prior invention cannot be raised — those are available only in district court or, within nine months of grant, in post-grant review.
This constrains search strategy. A validity search intended to support an IPR petition should focus on documentary prior art, because a devastating public-use case is simply unavailable in that forum.
The strongest petitions rest on references the examiner never saw. Prior art already of record can be raised but faces the argument that the Office already considered it, and the PTAB's institution rate has fallen sharply — from roughly 65% in October 2024 to about 37% by February 2026 — making the quality of the prior art more decisive than it was.
Finding prior art to invalidate a patent
Start from the claims, not the abstract. Identify the specific combination of elements that must be defeated, and note which are likely to be the point of novelty.
Read the file history first. It tells you what the examiner searched, what was cited, and which limitation was added to secure allowance. That limitation is usually where the case turns, and prior art disclosing it is worth more than prior art disclosing everything else.
Search the period immediately before the priority date. The most useful references cluster in the two or three years before filing, when others were working on the same problem.
Prioritise non-patent literature. Examiners search patents thoroughly and non-patent literature unevenly. That asymmetry is where the good references are.
Look for products, not just documents. A datasheet, a manual, an archived product page or a preserved specification can anticipate where no patent does.
Before relying on a prior art position
- Establish the effective filing date precisely, including any priority claims, because everything turns on it.
- Confirm the publication date of every reference with evidence, not assumption.
- Check enablement — an anticipating reference must enable a skilled person to make the invention.
- Read the file history to see what was already considered.
- Chart element by element, the same discipline as infringement analysis.
- Decide the forum before finalising the search, because an IPR restricts you to patents and printed publications.
- Do not confuse a novelty search with clearance — expired patents matter for one and not the other.