What Can (and Can't) Be Patented?

People often assume that any clever, original idea can be patented. It can’t. U.S. patent law draws some firm and occasionally surprising lines about what actually counts as an invention. Understanding those lines early saves inventors time and money — there is little point preparing a filing for something the law will never allow. This guide explains, in plain English, the four requirements every patent must satisfy, the categories courts have carved out as off-limits, and concrete examples on each side of the line. For the practical mechanics of filing, see our companion guide on how to patent an idea.

Requirement 1: Patent-eligible subject matter (§ 101)

The starting point is Section 101 of the patent statute, which permits a patent for “any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof.” Those four categories are the entire universe of patentable subject matter:

• A process is a method or series of steps — for example, a method of manufacturing a material or treating a disease.
• A machine is a device with moving or interacting parts, like an engine or a piece of equipment.
• A manufacture is an article made from raw or prepared materials, such as a tool or a molded product.
• A composition of matter is a chemical compound, mixture, or other combination of substances.

If your invention does not fit into one of these buckets, it is not patentable, no matter how valuable it is. (Designs and certain plants have their own separate patent tracks, but the core utility-patent system is built on these four categories.) Importantly, fitting a category is necessary but not sufficient — you still have to clear the next three hurdles.

Requirement 2: Useful (utility)

The statute requires an invention to be useful. In practice this is a low bar for most inventions: the invention has to have a specific, real-world use and has to actually work for that purpose. A new gadget that does something — even something trivial — generally satisfies utility.

Where utility bites is at the extremes. The classic example is the perpetual-motion machine: a device claimed to produce more energy than it consumes runs against the laws of physics, so it cannot operate as described. Because it doesn’t work, it isn’t “useful,” and the patent office will reject it. The same logic blocks inventions whose only asserted use is speculative or scientifically impossible.

Requirement 3: Novel (§ 102)

To be patentable, an invention must be novel — genuinely new. Section 102 says you cannot patent something that was already known to the public before you filed. The body of everything publicly available is called prior art: earlier patents, published applications, products on the market, academic papers, websites, and public demonstrations, anywhere in the world.

If a single piece of prior art already describes your invention, it is “anticipated” and not novel. A subtle trap here is that your own public disclosure can count against you. Publishing, selling, or publicly demonstrating your invention before filing can destroy novelty, which is why running a thorough check — and filing promptly — matters so much. We cover that process in our guide on doing a patent search before filing.

Requirement 4: Non-obvious (§ 103)

Even something new can fail the non-obviousness test in Section 103. The question is whether the invention would have been obvious, at the time it was made, to a hypothetical “person having ordinary skill in the art” — an imaginary expert who knows everything in the relevant field.

This is the requirement that trips up the most applications. You can’t simply combine two known things in a predictable way, swap one obvious material for another, or make a routine improvement and expect a patent. The invention has to reflect something more than an ordinary, expected step that a skilled professional would naturally take. Non-obviousness is judged against the prior art as a whole, not against any single reference, which makes it more demanding than novelty.

What courts have ruled off-limits: the judicial exceptions

Here is where many inventors are surprised. Even when something fits a Section 101 category and is useful, novel, and non-obvious, courts have decided that three kinds of subject matter still cannot be patented. These are the judicial exceptions, and they come from a line of Supreme Court cases — most famously Alice Corp. v. CLS Bank (2014), Mayo v. Prometheus (2012), and Association for Molecular Pathology v. Myriad Genetics (2013):

• Abstract ideas — including mathematical formulas and concepts, certain methods of organizing human activity, and mental processes you could perform in your head.
• Laws of nature — scientific principles and natural relationships, like a correlation between a drug dosage and a patient’s metabolism.
• Natural phenomena — naturally occurring things and “products of nature.” In Myriad, the Supreme Court held that a naturally occurring DNA segment is a product of nature and not patentable merely because it has been isolated.

The Court’s reasoning is that these are “the basic tools of scientific and technological work.” Letting one person monopolize a law of nature or a mathematical truth would block, rather than promote, innovation. To decide whether a claim falls into one of these exceptions, the patent office and courts apply a two-step Alice/Mayo framework: first ask whether the claim is directed to an exception, and if so, whether it adds enough beyond the exception to amount to a genuine “practical application” rather than just the abstract idea or natural law itself.

Concrete examples: patentable vs. not patentable

It helps to see both sides side by side.

Generally patentable:

• A new and useful machine, such as a redesigned engine component with a real performance gain.
• A new chemical compound or pharmaceutical formulation (a composition of matter).
• A novel manufacturing process with specific, non-obvious steps.
• A specific technical improvement to how a computer operates — not just an idea run “on a computer,” but a concrete advance in the technology itself.

Generally not patentable:

• A bare idea — “an app that helps people split bills” with no concrete, described invention behind it.
• A mathematical formula or algorithm standing alone (an abstract idea).
• A discovery of something in nature — a newly found mineral, plant compound, or gene sequence as it exists in nature (a natural phenomenon / product of nature).
• A perpetual-motion machine, which fails the utility requirement because it cannot work.
• Something already known (fails novelty) or an obvious tweak to existing technology (fails non-obviousness).

Software and apps sit in a notoriously gray zone, because many are challenged as abstract ideas under Alice. They are not automatically barred, but they need a genuine technical improvement to survive. We walk through that nuance in our guide on whether you can patent an app.

The “you can’t patent a bare idea” misconception

The single most common misunderstanding is the belief that you can patent an idea. You cannot. A patent does not protect a concept, a goal, or a “wouldn’t it be great if…” — it protects a specific, fully described invention: how it is built and how it works.

The legal requirement behind this is enablement: your application has to describe the invention in enough detail that a skilled person could actually make and use it without undue experimentation. An idea you can’t yet explain at that level isn’t ready to patent. The takeaway is practical — develop and document the real, working solution, not just the inspiration behind it. Until the idea becomes a concrete invention, there is nothing for a patent to protect.

How novelty and non-obviousness are actually judged

Both novelty and non-obviousness are measured against prior art, but they ask different questions.

Novelty is a one-to-one comparison: does any single piece of prior art already contain every element of your invention? If yes, it’s anticipated and fails. If no single reference covers it, it clears the novelty hurdle.

Non-obviousness is a broader, combined comparison: even if no single reference shows your whole invention, would the differences between your invention and the prior art have been obvious to a skilled person who could look at multiple references together? Examiners often combine references and ask whether there was a reason to combine them and a reasonable expectation of success. This is why an invention can be new yet still rejected as obvious — the pieces existed and combining them was the expected next step.

Because these determinations turn heavily on the specific prior art and how claims are written, this is exactly where experienced patent attorneys add value. You can read deeper treatment of eligibility and obviousness disputes in our patent topic analyses.

The bottom line

To recap what can be patented: an invention must fit a statutory category (process, machine, manufacture, or composition of matter), and be useful, novel, and non-obvious — while steering clear of the judicial exceptions for abstract ideas, laws of nature, and natural phenomena. You cannot patent a bare idea, a math formula, a discovery of nature, or a perpetual-motion machine. If your invention clears those hurdles, your next steps are searching the prior art and preparing a careful application that describes a real, working invention rather than a concept.

This guide is general legal information for educational purposes, not legal advice, and does not create an attorney-client relationship. Patent eligibility and obviousness are fact-specific and the law continues to evolve. For guidance on your particular invention, consult a patent attorney or agent licensed to practice before the patent office in your jurisdiction.