Who Appeals (and Wins) Patent Infringement Cases?

Key Takeaways

Only a small fraction of patent infringement suits are ever appealed, and fewer still receive a reasoned merits decision, so Federal Circuit doctrine rests on a thin, highly selected slice of cases; PAE cases are less likely to be appealed than practicing-entity cases, suggesting their high filing volume is driven more by settlement leverage than by willingness to litigate through appeal; Patent owners file most appeals, yet accused infringers achieve higher success rates when they appeal, indicating more selective or stronger defense-side appeals; Practicing/product-company patentees outperform PAEs on appeal, so patentee identity and business model materially affect expected appellate outcomes and settlement valuations; Published precedential opinions overstate patentee success relative to the full docket because many defense-favorable results are resolved via nonprecedential opinions or Rule 36 affirmances; Empirically informed predictions about who appeals and who tends to win can improve risk assessment, forum choice, and negotiation strategy in patent litigation.

Dataset Description

The authors build a linked dataset connecting three components: (1) district court patent infringement suits from the Stanford NPE Litigation Database (over 70,000 cases filed in U.S. federal district courts, 2000–2021), with detailed coding of patentee type (e.g., practicing entities/product companies, various categories of non-practicing entities, universities, individuals); (2) the Federal Circuit Dataset Project, covering all appeals docketed at the U.S. Court of Appeals for the Federal Circuit and all resulting opinions and Rule 36 summary affirmances from 2007 onward; and (3) hand-coded linkages between district court cases and their ‘first’ associated Federal Circuit appeals. The main analysis window focuses on district court patent infringement cases filed from 2011–2016 and their associated Federal Circuit appeals docketed through 2021, producing several thousand matched first appeals and roughly two thousand appeals with merits decisions (written opinions or Rule 36 dispositions). The jurisdiction is exclusively U.S. federal courts (district courts plus the Federal Circuit).

Key Findings

The study shows that only a small share of patent infringement cases ever reach the merits on appeal. For cases filed in U.S. federal district courts between 2011 and 2016, only about 5–6% are appealed to the Federal Circuit, and only about 60% of those appeals result in a merits decision in the form of a written opinion or a Rule 36 affirmance. Many appeals are terminated by dismissal or settlement before a merits disposition. Appeal behavior is strongly patterned by patentee type. Patent assertion entity (PAE) cases, which make up a large proportion of district court filings, are underrepresented on appeal relative to their filing share, consistent with high settlement rates and a business model focused on extracting settlements rather than litigating through appeal. In contrast, cases brought by practicing/product companies appear on appeal more often than their share of trial filings. Across all categories, patent owners initiate most appeals, but accused infringers fare better when they are the appellants: the Federal Circuit affirms district court outcomes less often in defendant-initiated appeals than in patentee-initiated appeals, implying that defense-side appeals are more selective or more meritorious on average. The identity and business model of the patentee also correlate with appellate success. Practicing-company patentees win in full or in part substantially more often than PAE-type patentees, whose appeal success rates are markedly lower, even though once before the Federal Circuit, the formal mode of disposition (precedential opinion, nonprecedential opinion, or Rule 36) looks broadly similar across patentee categories. Finally, the authors highlight that precedential published opinions systematically overrepresent patentee wins relative to the full universe of outcomes. Because many affirmances, especially routine, defense-favorable outcomes, are resolved via nonprecedential opinions or Rule 36 summary affirmances, relying solely on published doctrine yields a plaintiff-friendlier picture than the actual distribution of appellate results. Together, these patterns underscore that who appeals and what kind of decision the court issues are non-random, and that appellate case law is produced by a narrow, selected subset of disputes shaped by party type, business model, and strategic settlement choices.

Summary

The article investigates the appellate layer of U.S. patent infringement litigation by linking district court cases to their subsequent appeals in the U.S. Court of Appeals for the Federal Circuit. Using the Stanford NPE Litigation Database for trial-level information and the Federal Circuit Dataset Project for appellate dockets and decisions, the authors construct a relational dataset that follows cases from initial filing to potential appeal and Federal Circuit disposition. This design allows them to ask not just who wins at the appellate level, but which kinds of parties and disputes are actually making law at the Federal Circuit.

Their first core finding concerns volume and selection: very few patent infringement suits are appealed, and even fewer culminate in a merits decision. For district court cases filed between 2011 and 2016, only about one in twenty proceed to appeal, and about 40% of those appeals end without a merits ruling due to settlement or dismissal. This means the body of Federal Circuit patent doctrine is built on a small, non-random sample of all disputes. The authors show that patent assertion entities (PAEs), non-practicing entities that acquire patents for assertion, are especially prone to dropping out before appeal. Although PAEs account for a large share of district court filings, they account for a much smaller share of appeals. Practicing/product companies, by contrast, are overrepresented in the appellate pool. This reinforces a picture of PAEs as focusing on settlement extraction rather than on litigating all the way through the appellate process.

Within the set of appeals that do occur, party posture and identity shape outcomes. Across the board, patent owners bring most appeals, which aligns with the intuition that patentees must prevail on multiple grounds in the district court and thus have more to challenge. Yet accused infringers who choose to appeal are more successful than patentees: the Federal Circuit affirms district court judgments less often when the defendant is the appellant. The study further shows that practicing/product-company patentees win on appeal substantially more often than PAE-type patent owners. PAE appeals are both less common and less successful, suggesting that PAEs have weaker cases by the time they reach appeal or face a more hostile reception from the court, or both. For practitioners, these patterns suggest that expected appellate value depends not only on legal issues but also on the patentee’s business model and identity.

The article also examines how the Federal Circuit chooses the form of its decisions. A large share of affirmances in patent infringement appeals, particularly in more routine or defense-favorable cases, are decided through nonprecedential opinions or Rule 36 summary affirmances with no written opinion. Because practicing-entity wins and patentee-favorable outcomes are more likely to appear in precedential opinions, the published case law presents a more patentee-friendly picture than the full universe of appellate decisions. This decision-form selection matters for both scholarship and advocacy: lawyers who rely solely on reported precedents may misestimate real-world win probabilities, and scholars may infer doctrinal trends from a skewed sample.

Overall, the study reframes Federal Circuit patent doctrine as the product of multiple layers of selection, who files, who settles, who appeals, and what kind of decision the court issues. The authors’ empirical portrait suggests that appeals are more common and more productive for practicing companies and for carefully selected defense-side challenges, whereas PAE disputes are more likely to resolve without shaping appellate law. For litigators and policymakers, this means that risk assessment, forum choice, and settlement strategy in patent cases should incorporate an empirically grounded understanding of these selection effects, rather than relying solely on published opinions that constitute the visible tip of the appellate iceberg.

How the Study Advances Empirical Understanding of Legal Outcomes

The study finds that legal outcomes in patent appeals exhibit a measurable structure characterized by strong selection effects, in which only a narrow slice of disputes reaches a merits decision at the Federal Circuit. The results indicate that decision environments are shaped by institutional regularities in which patentee identity and business models, such as the distinction between practicing entities and patent assertion entities, correlate with repeatable patterns of appellate success and settlement behavior. The analysis shows that the study’s methodological approach is consistent with Pre/Dicta’s emphasis on empirical, case-based analysis of legal outcomes and decision contexts, as a qualified and necessary component of strategic litigation practice at a high level.

Pre/Dicta Debuts Suite of New Capabilities, Further Expanding its Predictive Insights into Judicial Decision-Making