How the numbers are made
Every figure on Diagsheet comes from three public NHTSA files. Here's what's in them, what the site does with them, and where the numbers mislead.
On this page
The three files
Owner complaints
NHTSA's complaint file holds every safety-related defect complaint the agency has received since January 1, 1995. Owners file them on NHTSA's website, through its app, by phone or by letter, and NHTSA says a complaint goes into the public database once personally identifying information is removed. Each record carries the make, model and model year, the component the owner picked, the date NHTSA received it, the mileage at failure if one was given, whether there was a crash or a fire, how many people were hurt, and the owner's description. Diagsheet reads the files of complaints received from 2015 on.
Recalls
A recall starts with a defect or noncompliance report the manufacturer files with NHTSA under federal rules (Part 573). For each make, model and year, the site pulls NHTSA's list of recall campaigns: the campaign number, the date NHTSA received the report, the component, NHTSA's summary of the defect, the consequence and the remedy, plus NHTSA's "do not drive" advisory where one applies.
Investigations
NHTSA's Office of Defects Investigation publishes a file of its own investigations: the action number, the vehicles covered, the component, the dates it opened and closed, the recall it led to if any, and a summary. The two letters at the front of the number give the type, and the glossary explains each one.
Counting
One complaint is one count. NHTSA lists a complaint once for each component the owner named, so a single report about a stall that also lit up the dash can appear under both the engine and the electrical system. The site counts it once in the vehicle total and once in each system's total, which is why the system figures on a page add up to more than the vehicle figure.
Most model years need 100 complaints before they get a page. The 30 models the site started with (and their hybrid versions) get one from 40, and model years from 2026 on publish from 10, because a new model's early reports are the point of those pages.
How complaints become problem groups
A problem on a vehicle page, such as "Steering sticks, feels notchy or jerks", is a group of complaints that meet two tests. The complaint has to be filed under a particular NHTSA component, and its text has to contain one of a set of words. For sticky steering that's any steering complaint that says stick, notch, jerk, bind or catch.
There are 56 groups, and the same rules run on every vehicle. 21 narrower sheets sit inside them, like the backup camera sheet inside the infotainment group. A report on a narrower sheet is always on the wider one too, so the narrow sheets never inflate a vehicle's main list. A report can land in more than one group if it describes more than one problem.
A group needs 25 reports on a vehicle before it gets its own sheet there, and 8 reports that state a mileage before it gets a bar on the mileage track. Across the whole site, 53% of complaints land in at least one published group. The rest still count toward the vehicle and system totals; they just describe something no group covers, or a group that's too small on that car to publish.
How the groups are checked
The groups added in September 2026 each had to pass a reading test. A sample of the reports a group caught was read to see how many were really about that problem, and the bar was eight in ten. A group for radio volume problems was cut when only about six in ten of its reports were about the volume. Two came in at around seven in ten and were kept anyway, "Brake warning light stays on" and "Infotainment screen freezes or goes black", so read those two with that in mind. (The first groups the site launched with predate that test.)
Groups also have to be common enough to be worth a page on many cars. Candidates for clutch problems, power windows, tire pressure sensors and several others were dropped because too few vehicles had enough reports for a sheet.
A word pattern is still a blunt tool. A complaint about a sticking door can end up in the sticky steering group if it happens to mention both. That's why every symptom sheet shows owners' own words next to the count, and those are the best check on whether a group fits your problem.
Mileage
NHTSA's form has a field for the mileage at failure, but plenty of owners leave it blank and write the number into their description instead ("at 84,000 miles", "around 84k"). Where the field is empty, the site reads the first mileage written in the text and keeps it only if it falls between 500 and 400,000 miles. Between the two, 47% of the complaints on the site carry a usable mileage.
The bar for a problem covers the middle half of those mileages, from the 25th to the 75th percentile, with a tick at the median. So a quarter of owners hit it earlier than the bar starts and a quarter later than it ends. Reports without a mileage still count everywhere else; they just can't place a problem on the track.
Prices
Dollar figures come only from complaints that quote one, between $100 and $15,000. Owners quote what they were charged, what a shop quoted them, and sometimes what the car cost, so read the price strip as the spread of figures owners mentioned. It isn't a repair estimate.
Tick sheets
The phrases on a symptom sheet are word patterns too. Tick several and the count shows only the reports that match all of them. It's a way to see whether your combination of symptoms is common on this car, not a diagnosis.
What the counts can't tell you
A complaint count isn't a failure rate. NHTSA's files say how many owners filed, not how many cars are on the road, and nothing on this site divides by sales. A best-selling model will collect more complaints than a rare one even if it's the more reliable car. Compare a model year against other years of the same model, which is what the model pages do, before comparing it with a different model.
Older cars have had longer to collect complaints. A car that's been on the road for ten years has had ten years of chances to file, while a two-year-old has had two. Vehicle and symptom pages show raw counts. Model pages divide each year's complaints by its years on the road before calling a year worst or best, and they never call a year before 2016 one of the best, because the complaint files the site reads start with reports received in 2015 and those early counts run short.
Owners who file had a bad experience. Filing is voluntary. Most drivers with a trouble-free car never visit NHTSA's site, so every count leans toward failure. Anything that sends owners to the form, like a recall notice or a news story, can push the count up without any change in the cars.
A complaint is one owner's account. NHTSA reviews every complaint for signs of a defect trend and may contact an owner for receipts or photos, but it doesn't contact every complainant, and the published text is the owner's account as filed. The site shows it that way.
A recall on the same system isn't proof. A recall on a car's steering doesn't mean every steering complaint on that car is the recalled defect. Recall pages list complaints on each recalled system so you can compare, and whether your own car is covered depends on its VIN.
There's no magic number. NHTSA says there's no set number of complaints that opens an investigation. A high count on this site doesn't mean an investigation is coming, and a low one doesn't mean a problem is minor.
How often it updates
NHTSA publishes its files daily and the site rebuilds from them every day. Each page carries a line with the date of the newest report behind it and the date it was last checked. The date a page gives search engines comes from that newest report too, so a page only looks updated when something new was actually filed.