What § 101 actually asks
Section 101 is written as a grant, not a bar: any process, machine, manufacture or composition of matter may be patented. The difficulty is entirely judge-made. Courts have long held that laws of nature, natural phenomena and abstract ideas are not eligible, because permitting a patent on the building blocks of technology would impede rather than promote innovation. Everything hard about eligibility comes from drawing that line.
The USPTO does not ask examiners to apply the case law directly. It gives them a flowchart, and understanding examination practice means understanding the flowchart, because that is what an Office action will actually recite.
The flow examiners follow
Step 1 — is the claim in a statutory category? A process, machine, manufacture or composition of matter. This screens out claims to a signal per se, a human being, or a bare mental process untethered to any category. Most claims pass. This is the whole of Step 1.
Step 2A Prong One — does the claim recite a judicial exception? Abstract idea, law of nature, or natural phenomenon. For abstract ideas the Office works from enumerated groupings — mathematical concepts, certain methods of organizing human activity, and mental processes — rather than from an open-ended inquiry. A claim that recites none of them is eligible and the analysis stops.
Step 2A Prong Two — is the exception integrated into a practical application? This is where most eligible claims are found eligible, and it is the prong worth arguing hardest. The question is whether the claim as a whole applies the exception in a way that imposes a meaningful limit, most commonly by improving the functioning of a computer or another technology. Critically, this prong is evaluated without regard to whether the additional elements are conventional. That distinction is the single most useful thing to know about the current framework: an improvement can be built out of well-known components and still integrate the exception into a practical application.
Step 2B — does the claim amount to significantly more? Only reached if Prong Two fails. Here conventionality does matter, and the burden of showing that an element is well-understood, routine and conventional rests with the examiner, who must support it with evidence rather than assertion.
Where the arguments usually are
Most § 101 disputes are really disputes about one of three things.
Whether the claim recites an exception at all. Prong One is a recitation question, not a gist question. An examiner who paraphrases a claim up to a level of abstraction at which every claim recites an abstract idea has over-generalized, and the MPEP's own instruction is to identify the specific limitation that recites the exception.
Whether the specification supports a technical improvement. Prong Two arguments succeed on the strength of the specification's description of the problem in the art and how the claim solves it. An improvement asserted for the first time in a response, with nothing in the specification behind it, tends not to carry.
Whether the examiner has evidence for conventionality. At Step 2B the Office must support a finding that an element is well-understood, routine and conventional. That requirement is a real constraint and is frequently the most productive thing to traverse.
Guidance has moved since this revision
This is the area where the pinned corpus is least sufficient on its own. The MPEP text here is fixed at Revision 01.2024, and the Office has issued a stream of eligibility material since — an update addressed to AI inventions, further subject-matter-eligibility examples, a memorandum reminding examiners how to apply the framework, and a designated Board decision. The tracked documents are listed below and are worth reading alongside the sections rather than instead of them: they change how the framework is applied without changing the framework.
Reading order
If you are new to a § 101 rejection: read § 2106 for the overall flow, then § 2106.04 for Step 2A, then the subsection matching the specific exception the examiner named. § 2106.07 is the one to read when the rejection itself looks defective, because it sets out what the Office is required to articulate when making and maintaining the rejection.