Two of the three crash tests measure how well a vehicle was engineered. The third mostly measures how tall it is.
We matched NHTSA rollover results against measured dimensions for the same model and year. The relationship is close to a straight line.
| Average height | Average width | Height ÷ width | Rollover rating |
|---|---|---|---|
| 142 cm | 182 cm | 0.784 | 4.57 |
| 147 cm | 182 cm | 0.813 | 4.37 |
| 161 cm | 183 cm | 0.879 | 4.17 |
| 168 cm | 188 cm | 0.895 | 4.00 |
| 178 cm | 197 cm | 0.905 | 3.83 |
| 191 cm | 201 cm | 0.954 | 3.37 |
Every step up in height costs rating, without a single exception in the sequence. From the lowest band to the highest is 1.2 stars.
The ratio does the work, not the height
Height alone would be a crude predictor, because a tall vehicle that is also very wide is more stable than a tall narrow one. The third column is the real variable.
Rollover resistance depends on how far the centre of gravity sits above the ground compared with how far apart the wheels are. Widen the track and the vehicle has to tilt further before it passes the tipping point; raise the mass and it has to tilt less.
The table shows both effects. The widest band, at 201cm, is also the tallest — and its extra width partly offsets the height, which is why it still manages 3.37 rather than falling further.
What this means about the test
It means the rollover score is largely predictable from a tape measure, before the vehicle reaches the track.
That is not a criticism of the test. Physics is physics, and a rating that reports it accurately is doing its job. But it does change how the number should be read: a three-star rollover result on a full-size SUV is not evidence that the manufacturer cut corners. It is evidence that the vehicle is tall, which you could see in the showroom.
The comparison worth making is within a shape class rather than across the whole market. A tall SUV that beats other tall SUVs has done something; a low sedan beating an SUV has done nothing but be low.
Stability control does not appear in this number
The obvious objection is that electronic stability control has been mandatory since 2012 and dramatically reduces rollover crashes. Both parts of that are true, and neither shows up here.
Stability control works before the event — it detects a slide and brakes individual wheels to keep the vehicle pointed where the driver steered. It prevents the loss of control that precedes most rollovers.
The rating measures what happens once the vehicle is already travelling sideways, at which point the only variables are geometry and mass. The two are addressing different halves of the same crash, and the rating covers the half electronics cannot reach.
How to weigh it
Rollovers are a small fraction of crashes and a disproportionate share of deaths, because the roof is the least protective part of a car and occupants are more likely to be ejected.
So the number matters, but it matters differently depending on the driving. Motorway commuting in a straight line, and the frontal and side scores dominate. Rural roads, evasive manoeuvres, gravel, a loaded roof rack — the rollover figure earns more weight.
One thing it cannot tell you is whether the stability system on a specific used vehicle is working, or whether the roof has been repaired after a previous incident. Both live in the vehicle's history rather than in the rating, and a free VIN check is where that record sits.
Frequently asked questions
Why do SUVs get lower rollover ratings?
Because the rating follows geometry rather than engineering. Matching test results to measured dimensions, vehicles averaging 142cm tall score 4.57 stars for rollover while those averaging 191cm score 3.37. The governing number is height divided by width — a taller, narrower vehicle has a higher centre of gravity relative to its track.
Can a tall vehicle ever get five stars for rollover?
It is rare. Width helps — the widest vehicles in our data partly offset their height — but no amount of electronic stability control changes the static ratio the test measures. Stability control prevents the loss of control that leads to a rollover; the rating measures what happens once the vehicle is already sliding sideways.
Does a low rollover rating mean the car is dangerous?
It means one specific risk is higher, and that risk is real: rollover crashes are a small share of accidents and a large share of fatalities. But the same vehicle usually scores well on frontal and side impact because mass helps there. The rating describes a trade, not a verdict.
Should I avoid vehicles with three-star rollover ratings?
Not automatically, but know what you are choosing. If most of your driving is motorway miles in a straight line, the frontal and side scores matter more. If it involves rural roads, evasive manoeuvres or roof loads, the rollover number deserves more weight than the overall star rating gives it.
Denis Kataev is a serial solopreneur and the founder of ForCar. With 15+ years in software engineering and 10 years in SEO, he builds data-driven products end to end — backed by a sharp eye for design. At ForCar he mines proprietary vehicle datasets, turning raw numbers into buying advice you can actually trust.
Keep reading
All articles →
Sudden Acceleration Reports Tripled in 2010. Then They Faded.
Unintended acceleration is the most contested failure in motoring. The complaint record does not settle the argument, but it does show something clear: reports of it follow the news as much as they follow the cars.
Every 250kg Cuts Injury Claims by About 9%. Someone Pays.
The protective effect of mass is usually argued from anecdote. Put measured kerb weights against claim frequencies and it becomes a clean line — along with the cost, which lands on somebody else entirely.
25 Cars Lost a Crash Star Between Model Years
A car that scored five stars last year is assumed to score five this year. Usually true, and the exceptions are informative: they cluster in two specific tests and almost never in the third.
Mark ForCar as a preferred source and our data shows up first in your Google results. One click, no account.