What the numbers say about IPR outcomes through 2026 and why the quality of the prior art search so often decides them
1. Introduction
Inter partes review (IPR) is the U.S. Patent Trial and Appeal Board’s (PTAB) trial procedure for challenging the validity of a patent on prior-art grounds. Created by the America Invents Act of 2011, it was designed as a faster, cheaper alternative to district court litigation. For years it was also remarkably effective for petitioners: most petitions were instituted and most patents that reached a final written decision lost at least some claims.
The picture in 2025 and 2026 is more complicated. The institution rate has fallen sharply as the USPTO leadership has expanded discretionary denial, petition filings have dropped and the Director has taken a direct role in institution decisions. Yet once a case is instituted, the final outcomes have remained strongly favourable to petitioners. Anyone planning, defending or valuing an IPR needs to understand both halves of that story and the role of the invalidity search that sits underneath it.
This article reviews published statistics through the third quarter of fiscal year 2026, explains how to read them and analyses how the quality and timing of a prior art invalidity search influences the stages at which IPRs are won or lost.
Scope and data caution. This is a general informational article, not legal advice. Statistics come from the USPTO’s published trial statistics and from commentary and reports by practitioner and analytics firms, which use different methods (for example, counting IPRs and PGRs together, counting by petition, patent or claim, or excluding joined cases). Figures are not strictly comparable across sources. USPTO policy has been changing quickly; verify current rules and data before relying on any figure. Public data do not isolate the effect of search quality as a variable, so the analysis in Sections 5 to 8 is explanatory and not a statistical finding.
2. How an IPR Works: The Decision Points That Matter
- Who and when. Any person other than the patent owner may petition, but not if they have already filed a civil action challenging validity and generally not more than one year after being served with an infringement complaint (35 U.S.C. 315).
- What grounds. IPR is limited to novelty and obviousness (35 U.S.C. 102 and 103) based on prior art consisting of patents and printed publications (35 U.S.C. 311(b)). Evidence such as products and public use cannot be the basis of the ground, although it may be used as supporting evidence for what a printed publication teaches.
- Institution. The Board may institute only if the petition shows a reasonable likelihood that the petitioner would prevail with respect to at least one challenged claim (35 U.S.C. 314(a)). Following SAS Institute v. Iancu (2018), institution is all or nothing: if instituted, all challenged claims and all grounds are considered. The institution decision is generally not appealable (35 U.S.C. 314(d)).
- Trial and final written decision. A final written decision must issue within one year of institution, extendable by up to six months for good cause (35 U.S.C. 316(a)(11)). The petitioner must prove unpatentability by a preponderance of the evidence (35 U.S.C. 316(e)).
- Estoppel. After a final written decision, the petitioner is estopped from asserting in district court or the ITC any ground that it raised or reasonably could have raised in the IPR (35 U.S.C. 315(e)).
- Appeal. Final written decisions may be appealed to the Federal Circuit.
Two gates therefore determine the result: institution and, if instituted, the final written decision. Settlement, which resolves many instituted cases before a decision, is a third path. Each gate has different success rates and each is influenced differently by the quality of the invalidity search.
3. What the Statistics Show
3.1 Institution rates have fallen sharply
During fiscal years 2022 to 2024, the PTAB instituted roughly two-thirds of IPR petitions (66 to 68 percent, according to USPTO-based figures cited by commentators). Reports covering fiscal year 2026 put the institution rate at about 41 to 43 percent, with one account placing it at 43 percent through part of the year and another at 41 percent for the full year, which ended on 30 September 2026. Differences stem from the date ranges and the counting methods used.
Several developments contributed. In 2025, USPTO leadership introduced an interim process for discretionary denial and gave greater weight to considerations such as parallel district court litigation timing (the Fintiv framework), the age of the patent and what has been described as the patent owner’s “settled expectations”. Later in 2025, the Director began to make institution decisions personally and there are reports that the Director has vacated a number of earlier grants of institution. A proposed rule published in the early part of the 2026 fiscal year, described by law firm commentary as aimed at parallel and repeat challenges, adds to the uncertainty. Reports of the number of new petitions reflect the same shift: one source reports that only 97 IPR petitions were filed in the third calendar quarter of 2026, compared with 327 in the same period of 2025 and other commentary describes quarterly filings in 2026 reaching record lows. Challengers have also turned in greater numbers to ex parte reexamination.
3.2 Final written decisions remain favourable to petitioners
Among IPRs and post-grant reviews that reach a final written decision, the proportion in which all challenged claims were held unpatentable has been in the range of 60 to 70 percent and the proportion in which at least one claim was found unpatentable has been around 80 to 85 percent.
| Period (IPR and PGR final written decisions) | All claims unpatentable | Mixed outcome | All claims patentable | Total decisions |
| FY 2024 | 312 (about 70%) | 67 (about 15%) | 67 (about 15%) | 446 |
| FY 2025 | 250 (about 65%) | 75 (about 19%) | 62 (about 16%) | 387 |
| FY 2026, Oct 2025 to Mar 2026 | 149 (about 61%) | 46 (about 19%) | 49 (about 20%) | 244 |
Table 1. Outcomes of IPR and PGR final written decisions, based on figures reported from USPTO data (percentages computed here from the reported counts; see IPWatchdog, May 2026).
Figure 1. Share of final written decisions by outcome (computed from the counts in Table 1).
The trend within these numbers is modest: a slow decline in the “all unpatentable” share and a rise in the “all patentable” share, but the overwhelming majority of final decisions still cancel at least some claims.
3.3 Reading the numbers by denominator
Headline statistics can be misleading unless it is clear what the denominator is. The same year looks very different depending on whether you ask “what share of final decisions invalidate all claims?” or “what share of all petitions end that way?”.
| Measure (USPTO data) | Result |
| FY 2025, outcomes by petition (1,470 petitions reaching an outcome) | Institution denied: 342 (23%). Settled: 368 (25%). Final written decision, all claims unpatentable: 250 (17%). Mixed: 75 (5%). All claims patentable: 62 (4%). Remaining outcomes included dismissals, adverse-judgment requests and other denials. |
| FY 2026 through Q3 (Oct 2025 to Jun 2026), outcomes by claim (19,196 claims challenged) | Institution denied: 8,822 claims (46%). Found unpatentable in a final decision: 3,803 (20%). Found patentable: 1,608 (8%). Other outcomes (settled, disclaimed, dismissed): the remainder. |
| FY 2026 through Q3, same data | About 20% of challenged claims and about 47% of claims in instituted trials, were found unpatentable. |
Table 2. Outcomes by denominator (source: USPTO PTAB trial statistics, FY 2025 Q4 and FY 2026 Q3 outcome roundups; figures as reported).
Two observations follow. First, from the perspective of a petitioner who files a petition, the chance of an outright win (all claims cancelled by final decision) is far lower than the “60 to 70 percent” figure often quoted, because denials and settlements account for a large part of the outcomes. Second, once a claim is in an instituted trial, the chance that it is cancelled is high: roughly one in two claims in the FY 2026 data. This is why practitioners often say the institution decision is the most important moment in an IPR.
3.4 Technology differences
Sector effects exist. For example, one analytics report puts the institution rate for bio-pharma petitions at around 50 percent in the first quarter of 2026, up from around 43 percent in the last quarter of 2025, which suggests that the discretionary factors affect different technologies unevenly, in part because of the different litigation and regulatory patterns of those fields (for instance, the Hatch-Waxman process).
4. Why Success Rates Are Not Simply “Search Quality”
It is tempting to read IPR success rates as a direct measure of how good petitioners’ prior art searches are. They are not, for several reasons:
- Selection effects. Petitioners file selectively, choosing patents they believe are weak and cases where they have strong references. The 2025 to 2026 fall in filings and the rise in denials mean that fewer and probably better-prepared, petitions get through.
- Discretion versus merits. A large and growing share of denials is based on discretionary factors (parallel litigation, patent age, prior challenges), not on the strength of the prior art. A superb search cannot overcome a discretionary denial.
- Settlement. About a quarter of petitions in FY 2025 ended in settlement and settlements are strongly influenced by the perceived strength of the petition, which does depend on the search.
- Panel and claim-construction effects. Outcomes also depend on claim construction, expert testimony, the persuasiveness of the motivation to combine and the quality of the patent owner’s response.
The invalidity search therefore matters as a necessary, not sufficient condition: a weak search very rarely wins, but a strong search can still lose on discretionary or procedural grounds. The remainder of this article examines the ways in which search quality shapes outcomes at each stage.
5. How the Invalidity Search Influences Each Stage
5.1 Before filing: deciding whether to file at all
Given the current institution environment, one of the most valuable outputs of a search is a decision not to file, or to file a different kind of proceeding. A search that produces only cumulative references, or references that were already before the examiner, signals a low chance of institution and a significant cost. A search that produces an anticipatory single reference, or a clear primary reference with an obvious secondary reference, supports a petition.
5.2 At institution: the “reasonable likelihood” threshold
The petition must show, claim by claim, a reasonable likelihood of prevailing. Search quality affects this in several ways:
- Novelty of the art to the Office (35 U.S.C. 325(d)). The Board may deny institution where the same or substantially the same prior art or arguments were previously presented to the Office, under the framework of Advanced Bionics v. MED-EL (2020). References that were not cited or considered during prosecution are the strongest answer to a 325(d) argument. A search that goes beyond the prosecution record, into foreign language documents, theses, standards, manuals and non-patent literature, is valuable here.
- Strength of the primary reference. A reference that discloses most claim elements, in the same field and context as the patent, makes the case simple. A weak primary reference forces a complex combination that is vulnerable to hindsight arguments.
- Coverage of dependent claims. Petitions that address only independent claims may leave a patent standing through dependent claims. Systematic claim-by-claim searching identifies references for the narrower limitations and this affects the claim-level success rates in Table 2.
- Evidence of public accessibility. A printed publication must have been publicly accessible before the critical date. The search must therefore capture the evidence (dates, indexing, library records, archive records, declarations) and not just the document. Petitions fail institution when public accessibility is not established.
- Effective date analysis. The search must be anchored to the correct priority or effective filing date for each claim. A reference that falls on the wrong side of the date, or a claim that is entitled to an earlier priority than the petitioner assumed, wastes the petition.
5.3 Timing and discretionary denial
Discretionary denial frameworks, including the Fintiv factors and the more recent considerations of settled expectations and prior challenges, depend heavily on timing and procedural posture. Early search and early filing of the petition, ideally well before the district court trial date, address those factors. Petitioners have also used stipulations, such as the Sotera stipulation (agreeing not to pursue in court any ground that was raised or could have been raised in the IPR), to mitigate the risk of denial, but the effect of such stipulations has changed with the USPTO’s current policy; they should be considered in light of current guidance. An invalidity search conducted at the start of a dispute and not after months of litigation, preserves these options.
5.4 After institution: carrying the burden at trial
Once the case is instituted, the petitioner bears the burden of persuasion by a preponderance. A thorough search gives the petitioner the following:
- Back-up grounds and reinforcing references that can answer the patent owner’s claim construction and amendment strategy.
- Evidence of the state of the art supporting the motivation to combine and reasonable expectation of success, which is often the contested ground in obviousness trials.
- Material to respond to secondary considerations, such as evidence that the claimed solution was known or that commercial success is attributable to features in the prior art.
- A foundation for expert testimony. Experts can only opine on the references that have been found and put in the record.
5.5 Estoppel and strategy
Because of statutory estoppel, an incomplete search is more dangerous in IPR than in other forums. If the petitioner loses, it may be barred from raising in court not only the art it used but art that a skilled searcher could reasonably have found. A thorough search both strengthens the petition and defines what can reasonably be argued to have been unavailable. Courts have differed on how broadly “reasonably could have raised” extends to art not actually found, so the consequence of a limited search should be assessed with counsel.
6. Anatomy of an IPR-Oriented Invalidity Search
- Understand and construe the claims. Identify the independent and dependent claims to be challenged and propose constructions under the standard used by the PTAB (the same as in district court, applying the ordinary and customary meaning in light of the specification and prosecution history). Searches built on the wrong construction find the wrong art.
- Fix the critical dates. Determine the effective filing date of each claim, including any priority or continuation issues and the dates for prior art under 35 U.S.C. 102(a)(1) and (a)(2).
- Review the file history. Study the references cited during prosecution, the examiner’s reasons for allowance, the applicant’s arguments and any prior challenges. This tells you what the Office has already seen and where it saw gaps.
- Run layered searches. Combine classification-based searches (CPC), keyword and Boolean searches, citation (forward and backward) searches, inventor and assignee searches and semantic or AI-assisted searches. Include non-patent literature, journals, conference proceedings, standards, textbooks, theses, datasheets, product manuals and archived web pages.
- Search across languages and jurisdictions. Foreign patent documents and non-English literature are frequently the best art and the least likely to have been considered.
- Screen and rank. Score references on how many claim elements they disclose, their date, their accessibility and how easily they combine.
- Chart the best references. Prepare claim charts for the top candidates, identifying single-reference anticipation possibilities and the best two- or three-reference obviousness combinations.
- Secure the evidence. Obtain authenticated copies, publication dates, archive captures and declarations that establish public accessibility.
- Test against likely responses. Anticipate the patent owner’s arguments, including claim construction, teaching away and lack of motivation and check whether additional art addresses them.
7. Linking Search Outcomes to IPR Results
| Search outcome | Likely effect at institution | Likely effect at final decision |
| Single reference anticipating the independent claims, not before the examiner | Strong; usually the best case for institution, though discretionary factors can still apply | High chance that all claims fall unless the patent owner amends or wins on claim construction |
| Strong primary reference plus a clear secondary reference with explicit motivation | Good, provided the motivation to combine is well supported | Good chance of success on independent claims; dependent claims need further art |
| Only cumulative art, or art already of record | Weak; exposed to denial under 325(d) | If instituted, the petition is vulnerable |
| Good art for independent claims but nothing for dependent claims | Institution possible | Mixed outcome likely; patent may survive in narrower form |
| Reference with unproven public accessibility | Risk of denial or being disregarded | May be excluded or given little weight |
| Broad search found nothing strong | Decision not to file avoids a loss and estoppel | Consider other tools (see Section 8) |
Table 3. How search results typically translate into IPR outcomes (qualitative guide, not statistical findings).
The 47 percent of instituted claims found unpatentable in FY 2026 (Table 2) hints at the effect of selection: petitions that clear a tighter institution gate tend to be the ones with strong art and strong art that is clearly charted tends to win. In other words, tougher institution standards raise the average quality of the instituted petitions and the search is the main source of that quality.
8. Strategy for Each Side
8.1 For petitioners
- Search early, before or at the start of litigation and decide on the forum with the results in hand.
- Invest in references the Office has not seen and build the petition around them, with explicit treatment of 325(d).
- Verify public accessibility and effective dates before filing.
- Chart dependent claims, not only independent claims.
- Consider the alternatives when institution risk is high: ex parte reexamination (35 U.S.C. 302 onward), which has a substantial new question of patentability standard, requires no estoppel and does not require the requester to participate; post-grant review for recently granted patents (within nine months of grant, with grounds beyond prior art); district court invalidity defences (clear and convincing evidence); and, outside the U.S., opposition proceedings (for example, within nine months of grant at the EPO and pre-grant and post-grant opposition and revocation actions in India under the Patents Act, 1970).
- Keep in mind the possibility of settlement; a well-documented search is leverage in negotiation.
8.2 For patent owners
- Commission your own invalidity search before enforcing or licensing and be ready to explain how the strongest art differs from the claims.
- Anticipate the discretionary arguments (parallel litigation, patent age, earlier challenges) as well as the merits.
- Make sure the prosecution record shows that significant art was considered, since this supports a 325(d) argument.
- Understand that surviving institution is not the same as winning: final written decisions in instituted trials have cancelled at least some claims in the large majority of cases.
8.3 For investors and licensees
Where a patent is central to a transaction, an independent invalidity search is a due diligence tool. Its results should be read together with the current institution environment: the lower the institution rate, the more a patent owner benefits from procedural protection, but the stronger the potential challenge based on art that has never been considered, which remains a risk through ex parte reexamination or in court.
9. Limitations and Open Questions
- Policy volatility. The USPTO’s institution rules and the Director’s role have changed repeatedly since 2025 and legal challenges to some of those changes may follow. Current guidance can change what the statistics mean.
- Data comparability. Sources count differently (by petition, patent or claim; IPR alone or IPR and PGR; fiscal or calendar year; with or without joined cases) and some of the figures cited here come from commentary on USPTO data rather than from the USPTO itself.
- Lagging indicators. Final written decisions in a given period come from petitions instituted a year or more earlier, under older institution practice, so the current post-institution statistics may not reflect the effect of the stricter institution gate. As the cohort changes, the final-decision numbers may move.
- Search effect is inferred. No public dataset isolates search quality as a variable; the conclusions in this article about the impact of searches rely on the structure of the proceedings and on practitioner experience.
10. Conclusion
IPR in 2026 is a harder gate and an unchanged trial. Institution rates have fallen from around two-thirds to roughly 41 to 43 percent and filings have dropped, but when a case is instituted, petitioners continue to win most of the time: around 61 percent of final decisions in the first half of fiscal year 2026 cancelled all challenged claims and about four in five cancelled at least one. In this environment the invalidity search is the single most controllable input. It decides whether to file, it determines whether the petition clears institution (including the 325(d) question), it provides the evidence and expert foundation for trial and it defines the estoppel risk if the petition fails. A search that is early, broad, documented and focused on references the Office has not seen gives a petitioner the best chance of success. And for patent owners, the same logic applies in reverse: understanding the strongest art before a challenge arrives is the best way to evaluate and defend a patent.
