Consumer Reports crash- tested rear-facing infant car seats at the speeds most cars are tested at and found that most of the seats failed disastrously. The findings are reported in the February 2007 issue.

Cars and car seats can't be sold in the U.S. unless they can adequately protect occupants in a 30-mph frontal crash. But most cars are also tested in the National Highway Traffic Safety Administration's (NHTSA) consumer information program in crashes at higher speeds, 35 mph for frontal crashes and 38-mph side crashes. Child car seats aren't.

Most of the infant seats failed when Consumer Reports crash-tested them at those higher speeds. The infant seats twisted violently or flew off their bases, in one case hurling a test dummy 30 feet across the lab. CR does, however, remind parents that any car seat is better than no seat at all.

All states and the District of Columbia require infants to be secured in car seats when traveling in passenger vehicles. Still, 572 infants under 1 year old were killed in traffic accidents from 2001 to 2005, with side crashes accounting for 151 of those deaths, or 26 percent, NHTSA data show.

Here are some highlights of CR's findings:

• Of 12 infant seats tested, only two performed well enough to be recommended by Consumer Reports: the Baby Trend Flex-Loc and the Graco SnugRide with EPS.
• Nine infant seats provided poor protection in some or all of the tests, even though they meet the federal safety standard. One seat, the Evenflo Discovery, didn't even meet that standard. CR is urging federal officials to order a recall of that seat.
• Many infant seats sold in Europe undergo more rigorous testing than do models sold in the U.S. Indeed, when CR crash-tested an infant seat purchased in England, the Britax Cosy Tot, it was the best in the tests. An infant seat sold in the U.S. by the same manufacturer, the Companion, failed CU's tests.
• CR's findings offer added evidence of problems with LATCH, the federally-mandated attachment system for child car seats. Many car seats performed worse with LATCH than with vehicle safety belts. And LATCH attachments aren't always easy to use.

Consumer Reports' new tests are tougher than the federal car-seat standard because a significant performance gap exists between vehicles and the car seats they carry.

"It's unconscionable that infant seats, which are designed to protect the most vulnerable children, aren't routinely tested the same as new cars," said Don Mays, senior director of Product Safety & Consumer Science for Consumer Reports. Consumers Union, the publisher of Consumer Reports, believes that the government should bring the safety testing for car seats in line with tests that are conducted on most new cars.

The federal New Car Assessment Program tests most cars and minivans, some pickups and SUVs, in 35-mph frontal crashes and 38-mph side crashes. Scores in the form of "star" ratings are widely publicized, and as a result carmakers have improved the crash protection of vehicles. There has been no such incentive for the makers of child car seats sold in the U.S.

In 2000, Congress mandated under the TREAD Act that NHTSA establish a consumer information program for child car seats incorporating ratings no later than November 2001. NHTSA concluded that the most effective consumer information system is one that gives the consumer a combination of information about ease of use and dynamic performance through higher-speed crash-test sled testing or an in-vehicle testing program. To date, NHTSA has not started providing dynamic crash protection ratings for car seats as part of its consumer information program. Currently, the agency's car seat information program includes ease-of-use ratings in the form of letter grades as well as tips and advice for parents.

The infant seats evaluated by CR are rear-facing carriers that snap in and out of a base. The base connects to the car by means of the vehicle's safety belts or LATCH attachments. (LATCH, which stands for Lower Anchors and Tethers for Children, includes belts that hook the base to metal anchors in the car.)

Consumer Reports crash-tested multiple samples of each infant seat. In some tests, CR used vehicle safety belts to secure the base; in other tests CR used LATCH attachments. The tests mimic a crash in a Ford Explorer SUV, a popular family vehicle. The Toyota Camry sedan crumples similarly, especially in a side crash, so CR would expect comparable results for some sedans.

CR used a test dummy weighing the maximum claimed weight for each seat. That's 30 pounds for the Graco SafeSeat and 22 pounds for the others.

In CR's 35-mph front-impact test, seven car seats failed. They separated from their bases, rotated too far, or would have inflicted grave injuries, as measured by CR's test dummy, whose sensors record the severity of impact. CR retested these to see whether they passed the 30-mph federal minimum standard. All passed except the Evenflo Discovery.

When Consumer Reports performed side-impact tests at 38 mph, eight models failed. Four of the seats flew out of their bases.

Three seats failed all of CR's more stringent crash tests: the Evenflo Discovery, the Graco Safe Seat, and the Britax Companion, formerly the top- rated seat based on earlier tests that mirrored the federal standard. Most other tested seats passed either the front- or side-crash test in some configuration, though only the Baby Trend Flex-Loc and the Graco SnugRide with EPS passed every test CR performed and therefore, garnered CR's recommendation.

Some Britax Companion seats were recalled in October because carriers were assembled incorrectly; CR tested a later model. The Evenflo Discovery, which CR is deeming Not Acceptable and believe should be recalled, was the subject of a NHTSA investigation in 2004 after the agency received seven reports about the carrier separating from its base. Evenflo received 52 reports, six involving fatalities, NHTSA says. The agency could not identify a safety defect and closed the investigation.

The Eddie Bauer Comfort infant seat also had problems, specifically in the fit-to-vehicle test. CR's trained safety-seat installers could not get one of two different bases supplied with the seat to fit securely when tested in five different vehicles. Because of that test result CR is judging the seat Not Acceptable and wants the problematic base to be recalled. The seat also performed poorly in the side-crash test when using the problematic base. The car seat (also called the Caress Comfort) is being discontinued though it is still being sold.

CR has learned that the manufacturer of the Eddie Bauer Comfort will supply an improved base through a "customer satisfaction program", but only to those consumers who know to complain to the company about improper fit. CR is crash-testing the seat with the better-fitting base. Results will be posted at www.ConsumerReports.org.