About
Car Problem Data is a public reference for vehicle defect records, published by Semalt.
Who publishes this
The site is published by Semalt, a search and web data analytics company that has worked with large public datasets since 2013. Editorial contact: [email protected].
We should be plain about where our competence lies. We are a data team, not automotive engineers, crash investigators or mechanics. We do not inspect vehicles, we do not test parts, and we do not form our own opinion on whether a car is safe. What we do is take federal files that are published in a form almost nobody can read and render them accurately and completely. Every judgement about a defect on this site belongs to the person who filed the complaint, or to the manufacturer that opened the recall.
Why it exists
Ask whether a used car is reliable and the search results hand you articles calling some component "notorious" with no number attached, or a forum thread with four owners in it. Meanwhile the evidence is sitting in public: NHTSA publishes every complaint an owner files and every recall a manufacturer opens. It ships as multi-gigabyte tab-separated text with no column headers, so in practice it goes unread.
We parse those files and publish what they contain — 1,569,818 complaints and 21,880 recall campaigns across 72,880 model years — so that a person deciding whether to buy a car, or worrying about a noise, can see the actual record.
Editorial standards
These are the rules we hold ourselves to, and you can hold us to them too.
- We publish the record, not a verdict. We never write that a vehicle is unsafe. We report how many people reported what, and let the counts speak.
- Complaints are labelled as allegations everywhere they appear. They are unverified, anyone can file one, and manufacturers dispute many. We say so on every page that shows them.
- Complaint text is reproduced as filed. We do not rewrite, summarise or sharpen an owner's words. Where NHTSA redacted something, the redaction stays.
- Our processing is documented in public. Every transformation between the source file and the page is written down in the methodology, including the data defects we found and the records we discard.
- Counts are reconciled against the source. Our figures match NHTSA's own public lookup. When they would not, we treat that as a bug in our pipeline, not an interesting finding.
- We do not accept payment to change what a record says. No manufacturer, dealer or law firm can buy the removal, softening or promotion of a defect record. If that ever changes for any part of the site, it will be disclosed on this page.
- We correct errors in public. See the corrections policy.
What this site is not
It is not a vehicle history report. We cannot tell you whether one particular car was crashed, flooded or had its odometer wound back — that information is not public. We show the record for the model year, not for an individual vehicle.
It is also not a substitute for a recall check. Recalls apply to specific production runs and their status changes as cars get repaired. For an individual car, use NHTSA's VIN lookup.
Car Problem Data is independent. It is not affiliated with, endorsed by, or acting on behalf of NHTSA, the U.S. Department of Transportation, or any vehicle manufacturer.
How current the data is
The records behind the site were built from the NHTSA flat files released in September 2026. The most recent complaint in our copy was filed on Sep 17, 2026, and the most recent recall report is dated Sep 14, 2026. NHTSA updates the source files continuously; we rebuild from them and stamp the currency on the data sources page.