A code names the test that failed. Not the part.
The codes, and which cars actually set them
1,138 codes with enough on file to be worth a page, built from 129,838 manufacturer filings and 5,819 owner reports. Not a dictionary — a record of where each code has turned up.
Why these pages are different
Every other page about a trouble code tells you what the code means. That is the easy half, and for a manufacturer-specific code it is not even knowable. The half nobody publishes is whether the code is a known pattern on the car in front of you — whether the manufacturer has already written a bulletin about it, how many owners of that exact model year reported it, and at what mileage. All of that is in public federal files. It has just never been joined up.
What a code is not
A stored code names the test that failed. It does not name the broken part, it is not a recall, and it is not a verdict on the vehicle. Read it as the first line of a diagnosis.
The codes with the most on file
Source: NHTSA filesEvery code with a page
Grouped by the system the first character names, most documented first, with a full index behind each one. The number beside a code is how many separate federal documents name it — manufacturer bulletins plus owner complaints.
Powertrain — 773 codes
The most documented first. See all 773 powertrain codes →
Body — 131 codes
The most documented first. See all 131 body codes →
Chassis — 86 codes
The most documented first. See all 86 chassis codes →
Network — 148 codes
The most documented first. See all 148 network codes →
The data behind this
Released under CC BY 4.0 — copy it, change it, build a product on it, including commercially. Credit WhichTrim and link back. See the carve-out in the terms.
Everything above, as files. No key, no form, no attribution beyond the line below — if it is useful, use it.
How to cite
WhichTrim, “Diagnostic trouble codes in the US federal vehicle record”, 2026-11-04. https://doi.org/10.5281/zenodo.22891283
Archived with a permanent identifier on Zenodo — that DOI always resolves to the current version, and the file is rebuilt nightly, so cite the DOI rather than a dated copy.
The underlying records are U.S. federal public-domain data from NHTSA. The joins, the filtering and the arithmetic are ours, and we would rather be credited than not — but the figures are checkable from the source either way, which is the point.
About these pages
Source: NHTSA files and the OBD-II standardWhere do these codes come from?
From two NHTSA files nobody reads together: manufacturer communications, where automakers name codes in the summaries of their own service bulletins, and the complaint database, where owners type in what a scan tool told them. We count both, per model year.
Why isn't every code here?
Because a page needs something to say. A code earns one when it has at least four mentions across two model years, or eight on a single one. Below that it still appears on the vehicle's own page, where the context makes it mean something.
Do you define manufacturer-specific codes?
No. The second character of a code decides who owns its meaning — a 1 or a 3 hands it to the automaker, so the same code can mean unrelated things on two different cars. Those pages carry the evidence instead, which is the part that transfers.
Is this a substitute for a scan tool?
No, and it is not a substitute for a diagnosis either. It tells you whether what your scan tool found is already a documented pattern on that nameplate-year, which is a question a scan tool cannot answer.
Check engine light codes: what an OBD-II code tells you, and what it doesn't
Since 1996, cars and light trucks have carried a federally required self-test system. How to read the code it stores, why that code names a failed test rather than a broken part, and why clearing it fixes nothing.