After years of study and debate, the federal government has finally come up with a way to make its rollover rating system more meaningful for consumers. Beginning with '04 models, the National Highway Traffic Safety Administration (NHTSA) will include results from test track maneuvers in gauging a vehicle's tendency to roll over.

NHTSA began posting rollover ratings three years ago in response to the alarming rise in deaths and injuries caused by this type of mishap, especially in sport-utility vehicles and pickups. Until now, however, government investigators based the ratings solely on a mathematical formula that accounts for a vehicle's track width and the height of its center of gravity.

Mere arithmetic didn't satisfy the system's critics, however. They claimed that NHTSA needed to include so-called "dynamic testing"-judging the vehicle's behavior during over-the-road maneuvers. Ever since, NHTSA has tried to develop a valid, objective way to measure rollover tendency while a vehicle is moving.

Now government investigators say they have a valid test: two tight turns (known as a "fishhook") at varying speeds on a test track. A computerized steering system will control each test vehicle to eliminate possible variations in driver performance from test to test.

The results from the new dynamic tests will be combined with results from the old, strictly mathematical method to produce the ratings. The current five-star system won't change. It ranges from five stars (the best, indicating a rollover risk of 10 percent or less) to one star (the worst, indicating a risk greater than 40 percent).

NHTSA doesn't require carmakers to post rollover ratings on window stickers or dealers to make them available in showrooms.

To find rollover ratings when you shop, visit www.nhtsa.dot.gov/cars/testing/ncap.