How The Lane Check scores a model year

Every score on this site is one number: the share of a model year's NHTSA owner complaints that are about the engine, transmission or fuel system. This page explains exactly how it is computed, what the labels mean, how fresh the data is, and what it cannot tell you.

By Eyal Malka · November 5, 2026 · Data: NHTSA, updated October 3, 2026

Where the data comes from

Three public NHTSA services, and nothing else:

• Owner complaints (NHTSA's complaints database), fetched for the year, make and model.
• Recall campaigns (NHTSA's recalls database), fetched the same way.
• vPIC, NHTSA's VIN decoder, used only when you check a VIN, to find its year, make and model. A VIN that fails its check digit, or that vPIC flags as invalid, is rejected rather than scored.

NHTSA often files one model under several names (an F-150 under each cab style, for example). We query the plain name and its close variants and merge them, counting each complaint once by its NHTSA record number and each recall once by its campaign number. Variants that are a different powertrain, such as electric, hybrid or plug-in versions, are kept out, because this is a drivetrain measure and an electric car's complaints would distort a petrol car's score.

The drivetrain share

Each NHTSA complaint carries one or more component tags. A complaint counts as a drivetrain complaint if any of its tags contains POWER TRAIN, ENGINE or FUEL/PROPULSION. The share is:

drivetrain share = drivetrain complaints ÷ all complaints for that model year

It is a share rather than a count so that a car that sold in huge numbers is not marked down just for having more owners to complain. The count is still shown next to the share on every page, because 60% of 40 complaints and 60% of 2,000 are not the same evidence.

The recall-flood adjustment

A mass safety recall (airbag inflators, headlight campaigns) can bury a model year in complaints that say nothing about the engine or transmission, and push a real drivetrain problem below the line. So the share is computed twice:

• Unadjusted: drivetrain complaints ÷ all complaints.
• Adjusted: the same, with non-drivetrain complaints tagged AIR BAGS, EXTERIOR LIGHTING, SEAT BELTS or CHILD SEAT taken out of the denominator.

The score is whichever is higher. When the adjusted figure is more than 2 percentage points above the unadjusted one, the page says so and shows both.

What the labels mean

Drivetrain share	Label
60% or more	Very high complaint signal
40% to under 60%	High complaint signal
33% to under 40%	Elevated complaint signal
Under 33%	No elevated signal
Fewer than 25 complaints	Insufficient data (not scored)

A high complaint signal means that 40% or more of what owners of that model year reported to NHTSA was about the engine, transmission or fuel system. It describes a pattern across the model year. It does not mean any particular car has the problem, and a "no elevated signal" year can still contain a bad car.

When we do not score

• Fewer than 25 complaints. Too few to read a pattern from. These model years are listed in the index but get no page.
• Battery-electric cars. They have no engine, conventional transmission or fuel system, so the share would always read low and falsely reassure you. The real used-EV risk, battery wear, is not something this measures. Hybrids have an engine and are scored normally.
• No records under that name. If NHTSA has no complaints filed under a model's name for that year, we say so rather than show a zero.

"Worse than X% of the model years we track"

Each scored model year is ranked against every model year we track: currently 1,790 model years of 200 models, 2010 to 2023. The figure is the percentage of those with a lower drivetrain share; ties count as half, and it is held between 1 and 99. The tracked set is the same for the public pages and for VIN reports, so the two always agree.

How the cut-offs were checked

The cut-offs were first set against a small panel of eight vehicles with known outcomes. On September 13, 2026 they were checked again against 438 model years (56 models, 2011 to 2018), using 72 model years with a documented drivetrain problem and 60 with a clean reputation. The 40% line caught 70 of the 72 documented problems. The 60% tier was added then to separate the worst of them; 2 of the 60 clean-reputation years reach it. That is a modest, hand-built check, so treat the labels as experimental.

Recalls

Recall campaigns are fetched for the same year, make and model and counted once per campaign. They are not part of the score: a recall is the manufacturer acknowledging a defect, and the share is about what owners reported. Car pages list the engine, transmission and fuel-system campaigns first. Whether a specific car still has an open recall depends on its VIN; a model-year page cannot tell you that.

What the full report adds

The free model-year pages show the score, complaints by component, the years either side, one owner report and the recalls. A VIN report runs the same analysis and adds sections the public pages leave out:

• The mileage owners stated when the problem appeared, shown only when enough drivetrain complaints give one, and always with how many it rests on. It describes those complaints; it is not a prediction for any car.
• Repair-cost ranges for the component most complained about, on elevated model years, labelled as planning estimates.
• Cleaner alternatives: nearby years of the same car and similar cars with no elevated signal, offered only when the car in hand is elevated.
• Up to 15 owner reports, with reports that name the car's own engine size first. Reports whose text names a different year or model are left out.
• An engine-level count, when the car's own engine size is named in at least 3 drivetrain complaints: how many complaints name each engine size. It is a count, not a failure rate, because most complaints never name the engine.
• What to ask the seller and check on the test drive, and two cars side by side.

How fresh the data is

• A VIN report asks NHTSA at the moment you run it.
• The public model-year pages are generated offline from a saved copy of NHTSA's records and refreshed roughly monthly: a saved result is reused for up to 30 days, then fetched again. The current pages were built from records fetched on October 3, 2026; each page shows the date its own numbers were fetched.
• An NHTSA outage or timeout is never saved as "no complaints". That model year is simply tried again on the next run.

Owners can file a complaint at any time, so a model year's numbers can change between refreshes.

Limitations

• Complaints are self-reported and unverified. Anyone can file one, NHTSA does not confirm them, and we do not either. Some are wrong, duplicated or filed against the wrong car.
• Volume is biased. How many owners complain depends on how many cars were sold, how old they are, and how much publicity a problem got. A share softens the sales effect; it does not remove the others.
• Tags are coarse. The score relies on the component tags each NHTSA record carries. A transmission complaint tagged only as something else is missed, and a complaint tagged "engine" may be about something minor.
• NHTSA files complaints against the model, not the engine. A model year sold with several engines gets one score.
• It is not a mechanical inspection, not a vehicle history report, and not advice to buy or not buy. It describes a model year, not the car in front of you. Get an independent pre-purchase inspection.
• The cut-offs are experimental, checked against a small, hand-built set of known outcomes.

Found something that looks wrong? Email support@thelanecheck.com with the page link. About The Lane Check covers who built it and how it makes money.