Every patent application is assigned to an art unit — a group of USPTO examiners who handle one slice of technology. Which art unit yours lands in is one of the strongest signals of how the next three to five years will go. Across the 618 art units with at least 1,000 decided applications on record, the median art unit allows 66% of the applications it decides. The twenty units below allow 36% or less — and the hardest of them allows barely one in five.
This page ranks art units by their measured allowance rate: allowed applications as a share of decided applications (allowed plus abandoned), computed from public USPTO records. It is a report of what happened, not a prediction of what will happen to any application. Every art unit links to its page in our free examiner directory, where you can see the statistics for each individual examiner in the unit — no login, no paywall.
| Rank | Art unit | Tech center | Allowance rate | Decided applications | §101 rejection rate | §103 rejection rate | Avg. pendency (mo.) |
|---|---|---|---|---|---|---|---|
| 1 | AU 3622 | 3600 | 21.1% | 9,305 | 85.5% | 81.2% | 49.6 mo |
| 2 | AU 1611 | 1600 | 24.7% | 12,381 | 16.4% | 90.1% | 40.9 mo |
| 3 | AU 1757 | 1700 | 25.7% | 1,703 | 14.0% | 94.7% | 38.5 mo |
| 4 | AU 3623 | 3600 | 26.5% | 12,738 | 87.5% | 87.3% | 47.6 mo |
| 5 | AU 3689 | 3600 | 27.6% | 7,940 | 88.1% | 90.2% | 47.5 mo |
| 6 | AU 3624 | 3600 | 28.2% | 12,474 | 87.2% | 86.8% | 46.4 mo |
| 7 | AU 1789 | 1700 | 30.6% | 9,050 | 9.2% | 94.6% | 47.8 mo |
| 8 | AU 3692 | 3600 | 30.8% | 7,278 | 86.0% | 80.6% | 49.0 mo |
| 9 | AU 1617 | 1600 | 31.4% | 17,495 | 13.4% | 88.2% | 38.4 mo |
| 10 | AU 1614 | 1600 | 32.0% | 10,141 | 14.3% | 74.0% | 36.6 mo |
| 11 | AU 1794 | 1700 | 33.3% | 20,358 | 14.6% | 84.5% | 42.2 mo |
| 12 | AU 3769 | 3700 | 33.6% | 6,285 | 27.4% | 87.1% | 43.7 mo |
| 13 | AU 3626 | 3600 | 34.0% | 13,935 | 89.6% | 85.4% | 41.7 mo |
| 14 | AU 3621 | 3600 | 34.6% | 8,844 | 73.8% | 84.8% | 47.3 mo |
| 15 | AU 1619 | 1600 | 34.7% | 10,568 | 10.4% | 89.2% | 38.8 mo |
| 16 | AU 1631 | 1600 | 35.1% | 13,365 | 54.3% | 74.7% | 44.8 mo |
| 17 | AU 3688 | 3600 | 35.8% | 9,050 | 80.0% | 81.8% | 45.7 mo |
| 18 | AU 1787 | 1700 | 35.8% | 12,518 | 13.6% | 90.3% | 43.8 mo |
| 19 | AU 3686 | 3600 | 35.9% | 11,545 | 83.6% | 86.1% | 46.0 mo |
| 20 | AU 3695 | 3600 | 36.0% | 7,596 | 82.5% | 78.7% | 45.7 mo |
The rejection mix in an art unit is not a personality trait — it is the case law mapped onto the subject matter the unit examines. Read the § 101 and § 103 columns above together and the list splits into two families, each hard for a doctrinal reason.
Ten of the twenty are in Technology Center 3600 — business methods, e-commerce, and finance. These units examine claims that sit squarely on what Alice Corp. v. CLS Bank and its progeny treat as abstract ideas: fundamental economic practices and methods of organizing human activity, implemented on general-purpose computers. That is why their signature is § 101 — between 74% and 90% of applications in these units receive a subject-matter eligibility rejection, the highest concentrations at the USPTO. The instructive contrast is with the software units in TC 2100, where computing inventions are examined as engineering rather than as commerce: § 101 is common there too — the median TC 2100 unit rejects 46% of applications on eligibility — but claims framed as technical improvements survive it, and the median unit still allows 65%. Not one TC 2100 unit makes this list; its hardest allows 44%, twice the rate of AU 3622. Software does not doom an application under § 101. Software claims that read as a business practice do.
The other ten are chemistry and life sciences — five units in TC 1600 (biotech and organic chemistry) and four in TC 1700 (chemical and materials engineering), plus one medical-devices unit in TC 3700. Their § 101 rates are ordinary because their claims are directed to compositions, formulations, and apparatus — things eligibility doctrine rarely touches. Their § 103 rates are extreme for an equally structural reason: obviousness doctrine gives chemical examiners unusually sharp tools. Close structural analogs raise a prima facie case; claimed ranges that overlap the prior art are presumed obvious; and combining known actives, excipients, or process conditions is routinely framed as ordinary optimization under KSR. In AU 1757 and AU 1789, roughly nineteen of every twenty applications receive an obviousness rejection. Prosecution in these units is therefore evidence work — unexpected results, teaching away, declarations under Rule 132 — rather than eligibility argument.
One unit on the list is hard both ways, and it is the exception that proves the mapping: AU 1631, computational biology and bioinformatics. Its claims are life science by subject but algorithmic by form, so it draws eligibility rejections under the Mayo/Alice line at a rate no other life-sciences unit approaches — 54% — stacked on top of a 75% obviousness rate. Where the docket straddles two doctrines, applicants face both.
The practical point: "hard art unit" is not one problem. The same application strategy that survives AU 3622 would be aimed at the wrong statute entirely in AU 1611 — and knowing which fight your unit picks is the first strategic fact about your application.
First, confirm who is actually examining it — art-unit averages hide wide spreads between individual examiners, and the difference between an examiner allowing 15% and one allowing 55% inside the same unit matters more than the unit's headline number. You can look up your examiner free with your application number. Second, read the unit's own page: each one shows every published examiner's allowance rate, rejection profile, and appeal outcomes, so you and your counsel can ground strategy in the record you are actually facing rather than the folklore about it.
Allowance rate is allowed ÷ decided (allowed + abandoned), pooled per art unit from public USPTO records as published on September 6, 2026. Only art units with at least 1,000 decided applications are ranked — 618 of 643 qualify — so a unit with a handful of dispositions cannot top the list on noise. Rejection rates are the share of applications receiving at least one rejection under that statute. Pendency is the average months from filing to final disposition. Pending applications are excluded from every rate. These figures describe examination history; they do not predict the outcome of any application.
This page reflects USPTO data as last refreshed on September 6, 2026, and is updated on an annual cycle — figures may not reflect activity after that date. The statistics are compiled from public records on a best-efforts basis and are not guaranteed to be accurate, complete, or precise; USPTO source records themselves contain corrections and revisions over time. Do not rely on any figure here as the sole basis for a filing or prosecution decision.
Lynch LLP is not affiliated with, endorsed by, or sponsored by the United States Patent and Trademark Office. These rankings and statistics are derived from publicly available USPTO data and do not predict the outcome of any particular application. This page is for general informational purposes and is not legal advice; using it does not create an attorney-client relationship. Full disclaimers →
ATTORNEY ADVERTISING — Sean Lynch, Partner, Lynch LLP
Lynch LLP represents applicants in patent prosecution before the USPTO. If you're researching examiners, these are general resources on the firm's services.