After considering a number of alternatives, NHTSA has decided that the dynamic test will use the so-called "fishhook" maneouvre - a series of abrupt turns at varying speeds. A computerised steering system will be used in each test vehicle to maintain objectivity.
In 2002 10666 people were killed in rollover crashes, up 5% from 2001. Some 61% of all occupant fatalities in sport utility vehicles (SUVs, or 4x4s) and 45% of pickup truck deaths were the result of a rollover crash.
By contrast, 22% of passenger car fatalities in 2002 were the result of a rollover crash.
"Consumers need to consider rollover risk when they shop for a new vehicle," said NHTSA Administrator Jeffrey Runge, M.D.
"Our rating system will give them the information they need to make a wise choice."
NHTSA's current consumer programme rates rollover risk based on a vehicle's "static stability factor," which is an engineering calculation based on the track width (the distance between two wheels on the same axle) and the height of the centre of gravity above the road.
Starting with the 2004 model year, the rollover risk predictions will be based both on the vehicle?s static stability factor and its performance in the dynamic test.
The rollover rating system - one to five stars - remains unchanged. One star is for rollover risk greater than 40%; five stars, 10% or less.
The agency has been providing consumers with vehicle rollover ratings since the 2001 model year as part of its New Car Assessment Programme (NCAP).