Every 510(k) clearance rests on a comparison, and the predicate device is the thing being compared to. If you only learn one term about how moderate-risk devices reach the U.S. market, this is the one, because the entire substantial-equivalence argument is anchored to it. A predicate is a device already legally on the market that a new device points to and says, in effect, 'I am like this, in the ways that matter to safety and effectiveness.'

The FDA does not leave 'already on the market' to interpretation. The 510(k) summary regulation, 21 CFR 807.92, spells out exactly which devices can serve as predicates. It requires the summary to contain an identification of the legally marketed device to which the submitter claims equivalence, and it defines that legally marketed device in three concrete categories.

"A legally marketed device to which a new device may be compared for a determination regarding substantial equivalence is a device that was legally marketed prior to May 28, 1976, or a device which has been reclassified from class III to class II or I (the predicate), or a device which has been found to be substantially equivalent through the 510(k) premarket notification process."— 21 CFR 807.92(a)(3), source

That date, May 28, 1976, is the enactment of the Medical Device Amendments, the moment Congress gave the FDA premarket authority over devices. Devices in distribution before that date were 'grandfathered,' and they became the original reference points. Every predicate in use today traces back, directly or through a chain of prior clearances, to that starting line. When a 2024 device cites a predicate that was itself cleared in 2018 against a predicate cleared in 2009, it is a lineage of substantial-equivalence findings stretching back toward the 1976 baseline.

Why the choice of predicate decides so much

Because the predicate sets the comparison, choosing it is one of the most consequential decisions in a 510(k). The new device must share the predicate's intended use. It must also share the predicate's technological characteristics, or, if it differs technologically, the submitter must show those differences do not raise new questions of safety and effectiveness. Pick a predicate with the same intended use and similar technology, and the equivalence story is clean. Pick one that diverges in intended use, and the device may not be substantially equivalent at all, which can send the submission toward a 'not substantially equivalent' determination and, potentially, the De Novo pathway instead.

The predicate is identifiable in the public record. Each cleared device carries a K-number and a product code, and the product code ties the device to a regulation number and device class. A submitter searching for a predicate is, in practice, searching the FDA's 510(k) and classification databases for legally marketed devices in the same product code with a matching intended use. The openFDA 510(k) dataset alone holds more than 173,000 substantially-equivalent decisions, an enormous pool of potential predicates, each one a device that already cleared the same standard.

What a predicate establishes, and what it does not

It is worth being exact about the logic. A predicate does not prove the new device is safe and effective from first principles. It carries forward the regulatory reasoning that the predicate, and the chain behind it, already satisfied. The new device inherits the assurance by demonstrating it is enough like the predicate that the same controls apply. This is the efficiency at the core of the 510(k): the FDA does not re-litigate the safety of a well-understood device type every time a comparable product enters the market.

That logic also has limits the regulation respects. A predicate must itself be legally marketed; a device that was withdrawn for safety reasons is a poor and sometimes impermissible reference point. And the comparison only reaches as far as the predicate's intended use and characteristics, which is why a genuinely novel device with no suitable predicate cannot use the 510(k) at all and must instead seek a De Novo classification or a premarket approval. The predicate is the hinge: it is what makes a 510(k) possible, and its absence is what forces a device onto a different pathway.

So when a device says it is 'FDA cleared,' there is always a predicate behind it, named in the 510(k) summary, identifiable by K-number, and tied to a product code and regulation number. Reading that predicate, what it is, what it is for, and how the new device matched it, is the most direct way to understand exactly what the FDA found when it issued the clearance. The standard, set in 21 CFR 807.92(a)(3), is the same for all of them: equivalence to a legally marketed device, traceable back to the 1976 line.

Primary predicate, reference devices, and the limits of the comparison

In practice, a 510(k) often names a single primary predicate, the closest legally marketed device with the same intended use, that carries the core equivalence argument. A submission may also cite additional reference devices to support specific technological characteristics that the primary predicate does not fully cover. The regulation's requirement is the same throughout: every device pointed to must be legally marketed and must fall within the three categories in 807.92(a)(3). The structure matters because the FDA evaluates the comparison as a whole; a clean primary predicate with a matching intended use is the backbone, and reference devices fill in around it.

The predicate also bounds what a clearance can mean. Because the new device's authorization flows from its likeness to the predicate, the cleared device's intended use cannot exceed the predicate's intended use without raising new questions that defeat substantial equivalence. This is why a manufacturer cannot quietly broaden a device's claims by selecting a convenient predicate; a mismatch in intended use is one of the most common reasons a submission ends in a 'not substantially equivalent' determination. The predicate, in other words, is not just the starting point of the argument, it is also the ceiling on what the clearance covers. That dual role, foundation and boundary, is what makes predicate selection the most scrutinized choice in the entire 510(k).