Head restraints in dozens of sport utility vehicles, pickup trucks and minivans provided only poor or marginal protection from neck injuries in simulated crashes conducted by the insurance industry.

The test results released Tuesday found several SUVs had improved protections against whiplash injuries but gave poor marks to vehicles made by several leading automakers, including BMW AG, DaimlerChrysler AG, Ford Motor Co., General Motors Corp., Nissan Motor Co. and Toyota Motor Corp.

The Insurance Institute for Highway Safety said the simulated rear crashes at 20 miles per hour showed that many large vehicles fall short in protecting against neck injuries, which lead to 2 million insurance claims a year costing at least $8.5 billion.

"In stop-and-go commuter traffic, you're more likely to get in a rear-end collision than any other crash type," said Institute vice president David Zuby. "It's not a major feat of engineering to design seats and head restraints that afford good protection in these common crashes."

The institute evaluated 87 current vehicle models based on the geometric measurements of the head restraints and their performance in a crash simulation sled. Fifty-four of the vehicles were rated marginal or poor, the two lowest rankings, while a dozen received the second-highest score of acceptable. Twenty-one received the best rating of good.

Automakers said there were many ways to evaluate rear crash protection and their vehicles were designed to provide a high degree of safety.

"We feel our test procedures are good predictors of how well our seat/head restraints will protect occupants from neck injuries in the event of a rear impact," said Toyota spokesman Bill Kwong in an e-mail.

Several SUVs made progress — 17 of 59 SUVs from the 2007 model year received top ratings in the testing, compared with six of 44 SUVs tested in 2006.

The best performers among 2007 SUVs included: Acura MDX and RDX; Lincoln MKX, Ford Edge and Ford Freestyle; Honda CR-V, Element and Pilot; Hyundai Santa Fe; Jeep Grand Cherokee; Kia Sorento; Land Rover LR3; Mercedes M Class; Mitsubishi Outlander; Subaru B9 Tribeca and Forester, and Volvo XC90.

SUVs from the 2007 model year rated poorly were: BMW X3 and X5; Buick Rainier, Chevrolet TrailBlazer, GMC Envoy and Isuzu Ascender; Cadillac SRX; Chrysler Pacifica; Dodge Nitro; Ford Explorer and Mercury Mountaineer; Mitsubishi Endeavor; Hummer H3; Hyundai Tucson; Jeep Liberty; Kia Sportage; Lexus GX470 and RX; Nissan Xterra; Saab 9-7X; Suzuki XL7; Toyota 4Runner and Highlander.

In other vehicle categories, the 2007 Toyota Tundra was the only pickup to receive the top score. Three minivans received the highest marks: Ford Freestar, Hyundai Entourage and the Kia Sedona.

For pickups, the institute gave poor ratings to Chevrolet Silverado 1500 Classic and the GMC Sierra 1500 Classic; Dodge Ram 1500; Ford Ranger and Mazda B Series; Nissan Frontier and certain versions of Ford F-150, Dodge Dakota and Mitsubishi Raider.

Minivans scoring poorly were the Buick Terraza, Chevrolet Uplander and Saturn Relay; some versions of the Chrysler Town & Country and Dodge Caravan; and the Toyota Sienna.

Another group of SUVs, pickups and minivans received either the second-lowest score of marginal or the second-highest rating of acceptable. A complete listing was available on the institute's Web site.

Several automakers defended their methods of testing for rear crash protection.

DaimlerChrysler noted that many of the vehicles were designed before the Institute began conducting the tests. General Motors said head restraints are designed for a variety of driver sizes and the restraints "are part of the integrated approach to occupant protection in all GM vehicles."

Nissan said in a statement that it designs "all of our products to provide a high level of occupant safety in a wide range of real-world crashes, including rear-impact collisions."

The crash simulation sled replicates the forces in a stationary vehicle that is struck in the rear by a similar vehicle at 20 mph. Vehicles got a higher rating if the head restraint contacted the dummy's head quickly and the forces on the dummy's neck and the acceleration of the torso were low.

The tests also consider the height of the restraint and its horizontal distance behind the back of the head of an average-size man.