Patent-eligible subject matter is the threshold question of what kind of thing the law will let you patent at all, before anyone asks whether your invention is new or non-obvious. The statute, 35 U.S.C. § 101, is short and generous: any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement of one. The trouble is that the Supreme Court has read three unwritten exceptions into that text, and the test for applying them, the Alice/Mayo two-step, decides the fate of most software and diagnostic patents.
The exceptions are abstract ideas, laws of nature, and natural phenomena. Nothing in § 101 mentions them. They are judge-made, dating back well over a century, and they rest on a single rationale the Court repeats in every case: these are the basic tools of scientific and technological work, and monopolizing them would impede innovation rather than promote it.
People conflate § 101 with § 102 and § 103 constantly, and the distinction matters because they fail differently. Novelty and obviousness ask whether someone got there first. Eligibility asks whether the thing you claimed is the sort of thing patents are for.
That is why a genuinely brilliant, unquestionably new discovery can still be ineligible. Newton could not have patented the law of gravitation. Einstein could not have patented E=mc². The Mayo Court said exactly this, and it is the hardest part of the doctrine for inventors to accept: you can be the first person in human history to find something, and the answer is still no, because you found a fact about the world rather than making something.
The categories themselves are broad. Diamond v. Chakrabarty (1980) held a genetically engineered oil-eating bacterium to be eligible, with the famous line that Congress intended eligible subject matter to include “anything under the sun that is made by man.” That phrase gets overquoted. The operative words are made by man, and the Court has spent the years since Bilski v. Kappos (2010) enforcing them.
Mayo Collaborative Services v. Prometheus Laboratories (2012) built the framework and Alice Corp. v. CLS Bank International (2014) generalized it to all subject matter, which is why practitioners name it after both. It runs in two steps:
Step one: Is the claim directed to a judicial exception? If no, the analysis ends and the claim is eligible. If yes, continue.
Step two: Do the claim elements, individually and as an ordered combination, contain an inventive concept sufficient to transform the exception into a patent-eligible application? The elements must amount to significantly more than well-understood, routine, conventional activity already known in the field.
Alice itself concerned computerized intermediated settlement, a way of using a third party to mitigate risk in financial exchange. Step one: intermediated settlement is an abstract idea, a fundamental economic practice long prevalent in commerce. Step two: reciting a generic computer to carry it out adds nothing, because doing something conventional faster with a computer is not an inventive concept. The claims fell, and thousands of business-method and software patents fell behind them.
Step one is where most of the real work happens, and it is the least rule-like part of the doctrine. Every claim, at some level of abstraction, is about an idea. The Federal Circuit acknowledged in Enfish, LLC v. Microsoft Corp. (2016) that describing claims at a high enough level “untethered from the language of the claims all but ensures that the exceptions to § 101 swallow the rule.”
Enfish is the leading escape hatch. The claims covered a self-referential database table, and the court held they were not directed to an abstract idea at all because they were directed to an improvement in the functioning of the computer itself. That framing, technical improvement versus generic automation, is now the primary dividing line for software claims.
DDR Holdings, LLC v. Hotels.com (2014) got there differently. The claims solved a problem that did not exist outside the internet, namely retaining visitors when they click a merchant’s ad on a host site. Because the solution was necessarily rooted in computer technology to overcome a problem specific to computer networks, the claims survived. Contrast the long line of cases invalidating claims that take an ordinary practice, add “on a computer,” and file.
Compare Diamond v. Diehr (1981), decided decades before Alice but still good law and still cited: a rubber-curing process that used the Arrhenius equation was eligible, because the claim was to an industrial process that happened to use a formula, not to the formula.
Step two turns on what was “well-understood, routine, conventional.” For years, courts treated that as a legal question they could resolve on the pleadings, which made § 101 a cheap and devastating early motion. A defendant could kill a patent on a motion to dismiss before any discovery.
Berkheimer v. HP Inc. (Fed. Cir. 2018) narrowed that. The court held that whether claim elements are well-understood, routine, and conventional is an underlying question of fact, and where the specification describes the elements as unconventional, that factual dispute defeats summary judgment. Aatrix Software v. Green Shades Software, decided days later, applied the same reasoning to motions to dismiss. The practical effect was to raise the cost of an early § 101 attack, not to change what is eligible.
The life-sciences side is, if anything, harsher. Mayo invalidated claims to a method of calibrating thiopurine drug dosing by measuring metabolite levels, because the correlation between metabolite concentration and efficacy is a law of nature and the surrounding steps (“administer,” “determine”) were conventional.
Association for Molecular Pathology v. Myriad Genetics (2013) held that isolated naturally occurring DNA is not eligible, since separating a gene from its surroundings does not create anything, while cDNA is eligible because it is synthesized and does not occur in nature.
The sharpest illustration is Ariosa Diagnostics v. Sequenom (Fed. Cir. 2015). Sequenom’s inventors discovered cell-free fetal DNA in maternal blood plasma and built a noninvasive prenatal test on it, replacing amniocentesis. The panel invalidated the claims, and Judge Linn concurred only because he considered himself bound by the sweeping language of Mayo. When the Federal Circuit denied rehearing en banc later that year, several more judges wrote separately to say much the same thing. That is the cleanest evidence that the doctrine reaches inventions no one thinks should be unpatentable.
The USPTO’s 2019 Revised Patent Subject Matter Eligibility Guidance restructured step one into a two-prong inquiry: first identify whether the claim recites an enumerated grouping (mathematical concepts, certain methods of organizing human activity, mental processes), then ask whether the claim integrates it into a practical application. If it does, the claim is eligible and examination never reaches step two. The Office layered a 2024 update on top addressing artificial intelligence claims.
This is more permissive than the case law, and the Federal Circuit has said plainly that the Office’s guidance does not bind it. The result is a two-track system: a claim can be allowed by an examiner following the guidance and later invalidated by a court that is not.
Nearly everyone agrees § 101 is unpredictable. Legislative repair has been proposed repeatedly, most prominently the Patent Eligibility Restoration Act, which would abolish the judicial exceptions and replace them with a short statutory exclusion list. It has been introduced across multiple Congresses and has not passed. Until it does, the Alice/Mayo two-step is the law, and the answer to any eligibility question begins by identifying the exception and hunting for the inventive concept.
What is patent-eligible subject matter? Under 35 U.S.C. § 101, a patent may issue on any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement of one. Courts have layered three judicial exceptions on top of that text: abstract ideas, laws of nature, and natural phenomena cannot be patented no matter how novel or useful they are. Eligibility asks what kind of thing you claimed, not whether it is new.
What is the Alice test? The Alice test, more precisely the Alice/Mayo framework from Alice Corp. v. CLS Bank (2014) and Mayo v. Prometheus (2012), is a two-step inquiry. Step one asks whether the claim is directed to an abstract idea, law of nature, or natural phenomenon. If it is, step two asks whether the claim’s remaining elements add an inventive concept that transforms the exception into something more than well-understood, routine, conventional activity.
Are software patents still eligible after Alice? Yes, but the claim has to do more than run a known business practice on a generic computer. Claims survive when they improve the functioning of the computer itself or solve a problem rooted in the technology. Enfish v. Microsoft (2016) upheld a self-referential database table as a technical improvement, and DDR Holdings v. Hotels.com (2014) upheld claims addressing a problem specific to networks. Claims that merely automate a human task usually fail.
Can you patent a gene or a diagnostic method? Isolated natural DNA is not eligible after Association for Molecular Pathology v. Myriad Genetics (2013), because isolating a gene does not change what nature made. cDNA is eligible because it is not naturally occurring. Diagnostic claims that detect a natural correlation and then apply conventional steps generally fail under Mayo, which is why Ariosa v. Sequenom invalidated a well-regarded prenatal test.
Going further: What Is Patentable, a screening checklist .
This page is general legal information, not legal advice, and it does not create an attorney-client relationship.
The Federal Circuit held a driveshaft manufacturing method ineligible under Section 101 for invoking Hooke's law, pushing Mayo deep into the mechanical arts.
The Federal Circuit held that specific rules for automating 3-D lip synchronization were a technological improvement, not an abstract idea under Section 101.
The Federal Circuit's first precedential machine-learning eligibility ruling holds that applying off-the-shelf models to a new data environment claims an abstract idea under § 101.
The Supreme Court held that a genetically engineered, oil-eating bacterium is patentable subject matter, opening the door to modern biotechnology patents.
The Supreme Court held that isolated human genes are unpatentable products of nature, while synthetic cDNA can be eligible because it is not naturally occurring.
The Federal Circuit held that whether a claim element is well-understood, routine, and conventional is a factual question that can defeat summary judgment of ineligibility under Section 101.
The Supreme Court held that a rubber-curing process is not unpatentable merely because it uses a mathematical formula and a programmed computer to control the cure.
A unanimous Supreme Court held that implementing an abstract idea on a generic computer adds nothing patentable, extending the Mayo framework to software and business methods.
The Federal Circuit held that a prenatal test using cell-free fetal DNA was patent-ineligible because it detected a natural phenomenon with conventional techniques, even as judges questioned the result.
The Federal Circuit upheld internet-centric claims that were necessarily rooted in computer technology to solve a problem unique to online networks, the first eligible software claims after Alice.
The Supreme Court held that a formula for updating alarm limits is patent-ineligible where the only novelty is the algorithm and the rest is conventional post-solution activity.
The Federal Circuit held that a self-referential database design was not an abstract idea, establishing that Alice step one is a meaningful filter for claims that improve computer functionality.
The Supreme Court held that combining naturally non-inhibitive strains of nitrogen-fixing bacteria was an unpatentable discovery of a product and phenomenon of nature.
The Supreme Court held that a method for converting binary-coded decimal numerals into pure binary is an unpatentable abstract idea because a patent would preempt the formula itself.
A unanimous Supreme Court held that correlations between drug-metabolite levels and dosing were unpatentable laws of nature, and built the two-step test that would reshape Section 101.
The Supreme Court rejected hedging as an unpatentable abstract idea while refusing to make the machine-or-transformation test the exclusive gatekeeper for process patents.
