Most bumpers on midsize cars do little to resist damage in the kinds of low-speed collisions that are common in commuter traffic and parking lots. The Insurance Institute for Highway Safety is releasing the first results of new tests that assess how well bumpers protect vehicles from expensive damage in everyday fender-benders.

Only 3 midsize cars among 17 the Institute tested -- the Mitsubishi Galant, Toyota Camry, and Mazda 6 -- withstood 4 bumper tests with $1,500 damage or less in each test. Some cars sustained more than $4,500 damage in just 1 of the 4 tests, and 2 cars rang up more than $9,000 total damage.

"Our tests measure how well bumpers protect cars from damage in everyday bumps," says Institute president Adrian Lund. "The whole purpose of bumpers is to keep damage away from headlights, hoods, and other parts that are expensive to repair, but this purpose was accomplished in only 2 of the 68 tests we conducted. In the rest, what we found is that bumpers aren't up to the job."

The new tests reflect the kinds of front and rear impacts that are common in the real world. Insurance claims of $4,500 or less for damage in these crashes total more than $6 billion each year.

Old versus new bumper tests: The Institute began conducting low-speed crash tests at 5 mph into a flat barrier in 1969. These tests led to the first federal bumper rules for cars, which required the bumpers to resist damage in impacts up to 5 mph. These requirements eventually were rolled back by the Reagan Administration in 1982. But recent research shows that some of the most costly low-speed crash damage occurs when vehicle bumpers slide under or over each other. This happens because the bumpers on colliding vehicles don't line up, and braking before the impact can lower the front end of a striking vehicle just before it hits the other vehicle. Under- and override often result in damage to vehicle grilles, headlights, hoods, and fenders.

The Institute's old flat-barrier tests were good indicators of bumper strength, but they didn't assess over- or underride. Vehicles with comparatively good performances in these tests still sustained costly damage in real collisions. The Institute's new series of tests comes closer to matching the damage that occurs in real-world impacts. Each car is run into a barrier designed to mimic the design of a car bumper. The steel barrier's plastic absorber and flexible cover simulate typical cars' energy absorbers and plastic bumper covers.

The four tests include front and rear full-width impacts at 6 mph and front and rear corner impacts at 3 mph. The barrier is 18 inches off the ground in the full-width tests and 16 inches from ground in the corner impacts. These heights are designed to drive bumper improvements and lead to better protection from damage in a range of real-world crashes. In developmental tests, these configurations produced the kinds and amounts of damage that commonly occur in low-speed collisions.

"We don't want the automakers to change bumper heights just to get good performance in our tests," Lund explains. "We want car bumpers to resist damage in real crashes with other cars as well as with higher-riding SUVs and pickups, so we revamped our tests to reflect such crashes. In particular, we want to encourage automakers to use bumpers with energy-absorbing bars that extend farther into vehicle corners to reduce damage to headlights and other critical and costly equipment. We want car bumpers that are taller so they engage the bumpers on SUVs and pickups instead of underriding them."

Bumpers still poorly designed: Many bumpers aren't high enough or tall enough to take the hit in crashes between cars and SUVs or pickups. Even when bumpers line up with those on other vehicles reasonably well, many don't stay engaged with the other bumpers in collisions or can't absorb the energy of even a minor bump. This means expensive car body parts sustain most of the damage.

"The cars with the lowest repair bills after our new bumper tests still sustained much more damage than they should have in some of the tests," Lund says. "We got crumpled grilles and headlights plus buckled fenders in impacts at speeds equivalent to an average person walking fast."

The full-front test represents a common situation where a car hits the rear of another vehicle that has stopped in traffic. In this test, the bumpers on only four cars -- the Galant, Camry, Mazda 6, and Saturn AURA -- stayed engaged with the test barrier instead of going under or over it. The result was lower damage totals than other cars in the same test. Damage to 3 of these 4 cars totaled less than $1,000, and the AURA was the only car among the 17 tested to limit damage to the bumper itself in the full-front test without getting into the car body.

"This test should be easy if cars had well-designed bumpers because the energy of the crash can be spread across the whole front of the car. Instead some cars sustained more damage in this test than the other three," Lund points out. The Nissan Maxima, Pontiac G6, and Volkswagen Passat each sustained more than $4,500 damage in the full-front test. Costly repairs were required to the cars' hoods, fenders, and headlights as well as air conditioning condensers.

"A big problem is that the Maxima, G6, and Passat underrode the barrier," Lund says. "Our research shows that this is also what happens in many real-world crashes. The bumper bar on the Passat, which is supposed to take the hit and absorb the energy in the crash test, wasn't even damaged. The car's grille, hood, fenders, and headlights were damaged instead."