The Insurance Institute for Highway Safety, in test results released Sunday, said that General Motors Corp. cars were among the worst performers, and it said that Volvo and Saab cars were among the best.
"We were disappointed with General Motors in particular because they have a lot of vehicles whose seats we didn't even bother to test, because the geometry wasn't good enough to warrant a dynamic test," Adrian Lund, the institute's chief operating officer, told CBS Radio News.
Altogether, 97 seat and headrest combinations found in 88 cars now on the market were studied by the institute, which tested 73 of those seats in a 20 mph rear-impact crash to see how well they would protect an average-size male dummy.
Eight seats earned the institute's highest rating, including those in the Volvo S40, S60 and S80, the Saab 9-2X and 9-3, and the Jaguar S-type.
"That's because these companies have been doing testing to see how their seats perform in an actual rear impact," said Lund.
Sixteen seats, including those in the Chevrolet Malibu and the Subaru Outback, got the second-highest rating of acceptable; 19 seats, including those in the Ford Focus and the Mini Cooper, earned the third-highest rating of marginal.
The other 30 received the institute's worst rating, poor, indicating the highest likelihood of neck injury in a rear-impact crash. Among those were the seats in the Audi A4 and S4, the BMW 3 Series, the Dodge Neon and the Jaguar X-type.
The institute did not test 24 seats — among them those in the Buick Regal, Cadillac Seville, Acura RL and Volkswagen Passat — because it determined the headrests were designed in a way that would not protect taller people.
In a statement, GM said it has been following the institute's guidelines for placement of headrests. The company also said it has led development of a headrest that moves according to the force of the crash.
GM said occupants come in many sizes and sit in various positions in the vehicle and cautioned against making changes based on one test.
"If the test methods chosen are not reflective of reducing real-world harm, there could be significant potential to cause seat design changes that are directionally wrong," GM said.
The institute has evaluated headrests for nearly a decade.
"What we're hoping what these new test results will do is to show people that those things that they can call 'headrests' are more than headrests, they are head restraints, and they are a safety feature," said Lund.
Neck injuries sustained in a rear-end crash rarely are life-threatening, but they happen frequently and can be painful and expensive.
"It's a fairly minor injury but it does cause a lot of pain and suffering, and it's very expensive: Over $7 billion a year in insurance injury claims," said Lund.
Whiplash happens when a vehicle is hit from behind and the seat propels forward. If the headrest does not move with the seat, the occupant's neck will bend back and stretch.
On the best-performing cars, the seat was sturdy but had enough cushion so the occupant's body could sink into it, keeping the head closer to the headrest. The headrests were positioned so that they would be close to the back of the head and protect tall occupants.
Some vehicles got different ratings depending on which seats were installed. For example, Chrysler Sebring with power reclining seats got an acceptable rating, but the Sebring with manual reclining seats was not tested because the institute determined its seats were inadequate.