35 U.S.C. § 101: Patentable Subject Matter

35 U.S.C. § 101 is the shortest and, for many university inventions, the most consequential of the four patentability sections. Where § 102 asks whether an invention is new and § 112 asks whether it is adequately described, § 101 asks a threshold question that comes before either: is this the kind of thing a patent can cover at all? For a technology transfer office (TTO), that question is not academic — it is the reason a promising diagnostic-method disclosure or a software invention can clear novelty and non-obviousness review and still be unpatentable, or can issue as a patent and then be invalidated years later in litigation on eligibility grounds alone.

This guide explains what § 101 actually says, the judicial exceptions courts have read into it, the Alice/Mayo two-step framework the USPTO now uses to apply it, and why diagnostic/biotech methods and software are the two invention types where university TTOs run into it most often.

What 35 U.S.C. § 101 actually says

35 U.S.C. § 101, titled “Inventions patentable,” reads in full:

Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.

On its face, this is an expansive grant: it names four broad statutory categories — process, machine, manufacture, and composition of matter — and Congress and the courts have long described the scope Congress intended for those categories as broad, covering “anything under the sun that is made by man.” Read in isolation, the text says almost nothing is categorically excluded.

What actually makes § 101 a significant obstacle in practice is not the statutory text itself but a body of Supreme Court case law that reads three judicial exceptions into it — exceptions that appear nowhere in the statute’s language but have been treated as implicit limits on patent eligibility for well over a century.

The three judicial exceptions

Courts have long held that three categories of subject matter, though not excluded by the statute’s text, are not eligible for patenting because they are treated as the basic tools of scientific and technological work — building blocks that no one should be able to remove from the public domain by patenting them outright:

Laws of nature — naturally occurring relationships, such as a correlation between a metabolite level in a patient’s blood and the correct drug dosage.
Natural phenomena — things that exist in nature independent of human invention, including naturally occurring DNA sequences.
Abstract ideas — including mathematical concepts, certain methods of organizing human activity (such as fundamental economic practices), and mental processes that can be performed in the human mind or with pen and paper.

A claim is not disqualified merely because it touches one of these categories somewhere in its scope — nearly every invention builds on a law of nature or a mathematical relationship at some level. The eligibility question is whether the claim, considered as a whole, is directed to the exception itself, or instead applies it in a way that adds something beyond the exception. Four Supreme Court decisions set the modern framework for that inquiry:

Bilski v. Kappos (2010) — held that a claimed method for hedging risk in commodities trading was an unpatentable abstract idea, and rejected the “machine-or-transformation” test as the sole test for process-claim eligibility.
Mayo Collaborative Services v. Prometheus Laboratories, Inc. (2012) — invalidated a diagnostic-method claim that told doctors to measure a drug metabolite and adjust dosage based on a stated correlation, holding that the correlation itself was an unpatentable law of nature and that the surrounding administration/measurement steps were routine, conventional activity that did not add an inventive concept. Mayo is the source of the two-step eligibility framework courts and the USPTO still use.
Association for Molecular Pathology v. Myriad Genetics, Inc. (2013) — held that a naturally occurring, isolated DNA segment is not patent eligible merely because it has been extracted from the body, but that synthetically created complementary DNA (cDNA), which does not occur in nature, can be eligible.
Alice Corp. v. CLS Bank International (2014) — extended Mayo’s two-step framework from laws of nature to abstract ideas, holding that a computer-implemented scheme for mitigating settlement risk was an abstract idea merely implemented on a generic computer, and that generic computer implementation does not by itself supply the “inventive concept” needed to make an abstract idea eligible. Alice is the case most software and business-method § 101 rejections trace back to, and gave the combined test its common name: the Alice/Mayo framework.

The Alice/Mayo two-step framework

The USPTO applies Alice/Mayo through a structured analysis, most recently organized in its 2019 Revised Patent Subject Matter Eligibility Guidance and codified in MPEP § 2106. It runs as follows:

Step 1 — statutory category. Is the claim directed to a process, machine, manufacture, or composition of matter? If not, it fails § 101 outright regardless of the exceptions below.
Step 2A, Prong One — does the claim recite a judicial exception? The 2019 guidance narrowed this inquiry by defining “abstract ideas” as falling into three enumerated groupings — mathematical concepts, certain methods of organizing human activity, and mental processes — rather than leaving examiners to compare a claim against prior case outcomes on an open-ended basis.
Step 2A, Prong Two — is the exception integrated into a practical application? If the claim recites a judicial exception, the next question is whether additional claim elements apply, rely on, or use the exception in a way that imposes a meaningful limit — for example, by improving the functioning of a computer or another technology, or by applying the exception to effect a particular treatment for a specific disease. A claim that passes Prong Two is eligible without reaching Step 2B.
Step 2B — is there an “inventive concept”? If the claim is still directed to a judicial exception after Prong Two, the final question (drawn directly from Mayo and Alice) is whether the additional elements, taken individually or as an ordered combination, amount to “significantly more” than the exception itself. Reciting well-understood, routine, and conventional activity — a generic computer performing generic functions, or a doctor performing a standard measurement — does not supply that inventive concept.

A claim survives § 101 by satisfying Step 1 and then either falling outside the judicial exceptions at Prong One, integrating the exception into a practical application at Prong Two, or supplying an inventive concept at Step 2B. Failing all of these routes at every claim in an application is what produces a § 101 rejection at the USPTO, or an eligibility challenge that can be raised at any point in litigation, including after a patent has already issued.

Where this bites hardest at universities: diagnostics and software

Two categories of university invention run into § 101 disproportionately often, for reasons that trace directly back to Mayo and Alice:

Diagnostic and biomarker methods. A common academic invention pattern — discover that a biomarker, gene variant, or metabolite level correlates with a disease state or drug response, then claim a method of “detecting X and thereby diagnosing Y” — is close to the exact fact pattern Mayo invalidated. Claims that recite only a natural correlation plus conventional detection steps (assays, sequencing, or measurement techniques that were already standard in the field) are vulnerable at Step 2B, because the correlation is the law of nature and the detection method supplies no inventive concept beyond routine lab technique. This has made diagnostic-method claiming one of the more difficult drafting problems in university biotech licensing since 2012, and it is a substantive reason TTOs increasingly steer diagnostic disclosures toward claiming a specific treatment step tied to the diagnostic result, a novel assay or reagent, or a non-naturally-occurring composition, rather than the correlation alone.
Software and computer-implemented inventions. Alice made abstract-idea rejections a routine part of prosecuting software claims, especially where the claim describes what a computer accomplishes (an outcome — organizing data, verifying a transaction, matching users) rather than a specific technical mechanism for accomplishing it. Claims that improve the functioning of the computer or network itself, or that solve a problem specific to computer technology, generally fare better at Prong Two than claims that simply automate a pre-existing manual or business process on a generic computer. This matters for university software disclosures spun out of AI, bioinformatics, and data-science research, where the underlying algorithm can itself look like an unpatentable mathematical concept unless the claim ties it to a concrete technical improvement.

By contrast, most claims for a new chemical compound, a new device, a new manufacturing process, or a synthetically modified biological molecule (like the cDNA at issue in Myriad) rarely raise § 101 problems, because they are not naturally occurring and are not abstract — the statutory categories and judicial exceptions were never really aimed at that kind of invention.

What this means for invention disclosure and claim strategy

§ 101 issues are best addressed early, before an application is drafted, not discovered for the first time in an office action:

Disclosure intake should flag eligibility risk. A TTO evaluating a diagnostic or software disclosure benefits from asking, at the same stage it screens for novelty, whether the core of the invention is a natural correlation or an algorithm, and whether the inventor can point to a specific technical application, device, treatment step, or non-naturally-occurring composition the claims can anchor to.
Claim drafting should build in a practical application, not just the discovery. Patent counsel drafting around Alice/Mayo typically works to include claim elements that tie the underlying correlation or algorithm to a specific, concrete use — administering a particular treatment based on a diagnostic result, or a technical improvement to how a system operates — rather than claiming the correlation or algorithm in the abstract.
§ 101 rejections can often be addressed, not just appealed. Because Step 2A Prong Two and Step 2B both turn on what the claim as a whole does beyond the exception, amending claims to add the practical-application language already present in the specification is a common, and often successful, response to a first-action § 101 rejection — which is part of why the quality of the original specification under § 112 matters for eligibility outcomes as well as for enablement.
Provisional filings are not exempt. A provisional application is never examined for eligibility, but if the practical-application detail needed to survive Alice/Mayo isn’t in the provisional’s disclosure, it cannot be added later without risking a loss of priority date for that material — another reason eligibility strategy needs to start at first filing, not at the non-provisional stage.

None of this makes diagnostic or software inventions unpatentable as a category — thousands issue every year. It means the drafting has to do more work to show the claim is directed to an application of the underlying discovery, not the discovery itself.

Frequently asked questions

What is 35 U.S.C. § 101?

It is the section of the U.S. Patent Act that defines what kinds of inventions are eligible for patent protection in the first place — process, machine, manufacture, or composition of matter — and, through Supreme Court case law, excludes laws of nature, natural phenomena, and abstract ideas from that eligibility unless a claim applies them in a way that adds an inventive concept or practical application.

How is § 101 different from § 102 and § 103?

§ 102 asks whether an invention is new (not already disclosed to the public). § 103 asks whether it is non-obvious over the prior art. § 101 is a threshold question that comes before either: is this even the kind of subject matter a patent can cover? An invention can be perfectly novel and non-obvious and still fail § 101 if it is directed to an unpatentable law of nature, natural phenomenon, or abstract idea.

Can software be patented?

Yes, but a computer-implemented invention has to clear the Alice/Mayo framework: claims that merely automate a pre-existing process on a generic computer are vulnerable as abstract ideas, while claims that improve the functioning of a computer or network, or solve a problem specific to computing technology, are much more likely to be eligible.

Can a diagnostic method be patented?

Diagnostic-method claims that recite nothing more than a natural correlation plus conventional detection steps are highly vulnerable after Mayo v. Prometheus. Claims that tie the diagnostic result to a specific treatment step, a novel assay, or a non-naturally-occurring reagent or composition have a substantially better eligibility profile.

Is isolated human DNA patentable?

No — under Association for Molecular Pathology v. Myriad Genetics, a naturally occurring DNA segment is not patent eligible merely because it has been isolated from the body. Synthetically created cDNA, which does not occur in nature, can still be eligible.

For the other statutory patentability requirements, see CASRAI’s guides to 35 U.S.C. § 102 (novelty) and 35 U.S.C. § 112 (specification and enablement). For how a disclosure moves from initial screening through filing, see the guides on provisional patent applications and the cost of filing a patent, and for software-specific licensing considerations, see open source software licensing in university technology transfer.