How do you calculate the speed at which a car rolled over?

From the distance it travelled rolling: v₀ = √(vf² + 2·μ·g·d). The factor μ is not road friction but the mean deceleration of the vehicle while rolling, which the literature places between 0.36 and 0.65 g.

How do you know whether a rollover was possible?

With the critical rollover speed for tripping, which depends on track width, the height of the centre of gravity and the roll inertia. It is a lower bound: below it a rollover is comfortably ruled out, above it a rollover is possible but not certain.

Which vehicles roll over most easily?

Those with a high centre of gravity relative to their track width. The measure that sums this up is the Static Stability Factor; below 1.05 the vehicle is in the lowest rollover-resistance group published by the NHTSA.

At what speed does a car roll over?

Updated 2026 · 5 min read

Two questions sound alike and are answered very differently: how fast the vehicle was going when it began to roll, and whether the rollover was possible at all given what is known about the case. The first is kinematics; the second is causation, and it is usually the one in dispute.

1 · The speed at the start of the roll

A vehicle that rolls covers a distance rolling until it stops, and that distance is measured reliably on site. The speed follows from it:

v₀ = √( v_f² + 2 · μ · g · d )

The delicate point is what μ is here. It is not road friction: it is the mean deceleration of the vehicle while rolling, and it covers wheel impacts, roof impacts and ploughing into the ground. The literature places it between 0.36 and 0.65 g, drawing on eleven separate sources.

The deceleration is not constant

Against real rollover tests, profiles are fitted that run from 0.78 g at the start to 0.29 g at the end. The value used is a mean valid for the complete rollover. Applying it to an isolated stretch —«only the first two rolls were measured»— produces a large error and is not defensible in a report: either the whole rollover is measured, or the result is declared as an order of magnitude.

And an important qualification about what comes out: it is the speed at the start of rolling, not the speed before control was lost. If there was a prior skid or yaw, that part is handled separately and the energies are added.

2 · Whether the rollover was possible

The other question arises when someone argues that the vehicle could not have rolled as described. It is answered with the critical rollover speed for tripping: the minimum lateral speed which, sliding sideways into a kerb, soft soil or a rut, is enough to tip it.

It depends on the geometry of the vehicle —track width and height of the centre of gravity— and on its roll inertia. From those a specific threshold follows.

How the threshold is interpreted

It is a lower bound: the classical formulation does not subtract the energy absorbed by the suspension and the tyres on tripping, so a real rollover requires somewhat more speed. Below the threshold, a tripped rollover is ruled out comfortably. Above it, it is possible, not certain. That asymmetry is what makes it useful: it serves far better to rule out than to assert.

Which vehicles roll first

The measure that sums it up is the Static Stability Factor, relating track width to the height of the centre of gravity. Below 1.05 the vehicle falls into the lowest rollover-resistance group published by the NHTSA — the territory of tall SUVs and vans. A low car sits well above that, and slides before it rolls.

Module 08 answers both questions: the speed from rolling distance and the critical tripping threshold, estimating the roll inertia from the vehicle geometry when it is not known.

Open the calculator →

Frequently asked questions

How do you calculate the speed at which a car rolled over?: From the distance it travelled rolling: v₀ = √(vf² + 2·μ·g·d). The factor μ is not road friction but the mean deceleration of the vehicle while rolling, which the literature places between 0.36 and 0.65 g.
Is the deceleration constant during a rollover?: No. Against real tests, profiles are fitted running from 0.78 g at the start to 0.29 g at the end. The value used is a mean valid for the complete rollover; applying it to a partial stretch produces a large error.
How do you know whether a rollover was possible?: With the critical rollover speed for tripping, which depends on track width, the height of the centre of gravity and the roll inertia. It is a lower bound: below it a rollover is comfortably ruled out, above it a rollover is possible but not certain.
Which vehicles roll over most easily?: Those with a high centre of gravity relative to their track width. The measure that sums this up is the Static Stability Factor; below 1.05 the vehicle is in the lowest rollover-resistance group published by the NHTSA.

Sources

The rollover calculation method and the stability and inertia data it relies on:

Rose, N. & Beauchamp, G. «Development of a Variable Deceleration Rate Approach to Rollover Crash Reconstruction». SAE 2009-01-0093, Kineticorp.
Heydinger, Bixel, Garrott, Pyne, Howe & Guenther (1999). «Measured Vehicle Inertial Parameters — NHTSA’s Data Through November 1998». SAE 1999-01-1336. Database of 496 vehicles measured at the VRTC’s Vehicle Inertia Measurement Facility.
Rollover Stability Measurements for NCAP. Annual reports produced by S-E-A Ltd. for NHTSA. They give each vehicle’s measured Static Stability Factor; the 2020 and 2021 model years add up to 90 vehicles with track width and centre-of-gravity height.

How these methods compare against documented cases is detailed in How this calculator is validated.