About RoadVitals

What this is

The U.S. government publishes an enormous amount of vehicle safety data, and almost none of it is pleasant to read. Complaints, recalls, defect investigations and manufacturer bulletins live in four separate NHTSA datasets that name the same vehicle four different ways; crash-test ratings sit behind a separate API; fuel economy comes from the EPA in a file of its own.

RoadVitals imports all of it, normalises the make and model names so one vehicle is one page, and puts a model year’s complete public record in one place: what owners have reported and about which components, which recall campaigns cover it, whether NHTSA has an investigation open, which bulletins the manufacturer has filed, how it performed in crash tests, and what the EPA says it consumes.

We also keep the history. Government sources publish only the current cumulative totals — there is no upstream record of what a vehicle looked like six months ago. We snapshot every vehicle’s aggregates daily, which is the only reason the trend charts and the recent-activity views on this site can exist at all.

Who it is for

• Owners checking what has been reported about a car they already have, and whether a recall campaign covers it.
• Buyers researching a specific model year before committing, who want the record rather than a review.
• Anyone comparing model years of the same vehicle to see how the reported picture changed between them.
• Journalists and researchers who want a structured view of NHTSA data with the source record one click away.

How we handle the data

Safety data is easy to misrepresent by accident, so the rules below are enforced in code, not left to judgement page by page.

• A complaint is a report, not a finding. Complaints are reports submitted to NHTSA by owners and drivers. NHTSA does not verify most reports before publishing them, and a complaint does not establish that a defect exists. Volume is influenced by how many of a vehicle were sold, how long it has been on the road, and how much attention an issue has received.
• Recalls, investigations, manufacturer bulletins and confirmed defects are four different things, and we never blur them together. Each has its own definition on the page it appears on.
• Every factual sentence about a vehicle is a template filled from database values. There is no generated prose about any vehicle anywhere on this site — no AI-written filler paragraphs, no invented narrative about what an issue “usually” means.
• We never fabricate a number. No estimated repair costs, no failure rates, no sales figures, no made-up reliability scores. Where we have no data, the page says we have no data.
• We withhold weak numbers rather than dress them up. Where there is too little evidence to compute a meaningful Issue Index, no score is shown at all — a printed figure gets used regardless of the caveat next to it.
• Our metric is ours. The Vehicle Issue Index is our own informational metric, not an official safety rating and not a reliability prediction. NHTSA’s 5-Star rating is an official government result and is always presented as theirs.

The methodology page documents the Vehicle Issue Index in full, including the popularity bias we cannot remove and the thresholds below which we publish nothing. Data sources lists every dataset, what we take from it, and how often it refreshes.

What we deliberately do not do

• Rank vehicles as “worst” or “most unreliable”. Complaint volume is driven by how many were sold and how long they have been on the road, so a leaderboard built on it would be a popularity chart wearing a lab coat.
• Tell you whether to buy, sell, keep or repair a vehicle. That is not our call.
• Present an owner’s report as an established fact about a model. We show complaints in the reporter’s own words and label them as exactly that.
• Publish a page just because a URL could exist. A vehicle with almost no records is still reachable, but it is not indexed and it is not padded out with filler.

Corrections

We will get things wrong — a name normalised into the wrong model, a record joined to the wrong year, a figure that does not match its source. If you spot one, email [email protected] with the page URL and the campaign or ODI number, and we will check it against the source record and fix our side.

One limit worth being clear about: if the government record itself is wrong, we cannot quietly edit our copy of it, because the next import would overwrite the edit and our page would no longer match the source anyone can verify. In that case the correction has to be made at NHTSA, and we will point you at the right form. To report a safety problem with a vehicle, use NHTSA’s reporting form — complaints filed there become part of the public record this site reads from.

Independence

Data is sourced from NHTSA and the U.S. Department of Energy / EPA and is presented for information only. We are not affiliated with any government agency or vehicle manufacturer.

If the site ever carries advertising, it will be labelled as advertising and it will have no influence on what the data says — there is no editorial layer for an advertiser to lean on, because every vehicle page is generated from the same records by the same rules.

Get in touch

General questions go to [email protected]. See contact for which address to use for what, and for the things we cannot help with — we cannot advise on an individual vehicle or a repair.