Filed at 9:44 p.m. ET

YONKERS, N.Y. (AP) -- Most of the infant car seats tested by Consumer Reports ''failed disastrously'' in crashes at speeds as low as 35 mph, the magazine reported Thursday.

The seats came off their bases or twisted in place, the report said. In one case, a test dummy was hurled 30 feet.

Of the 12 car seats tested, Consumer Reports said it could recommend only two, and it urged a federal recall of the poorest performing seat, the Evenflo Discovery.

Evenflo issued a statement disputing the tests' validity, saying, ''The magazine's test conditions and protocols appear to conflict with the collective experience of car seat manufacturers, NHTSA (the National Highway Traffic Safety Administration) and the scientific community.''

To be sold in the United States, an infant seat must perform adequately in a 30 mph frontal crash, and Consumer Reports found that all but the Discovery did so. But it noted that NHTSA crash tests most cars at higher speeds -- 35 mph for frontal crashes and 38 mph for side crashes -- so the magazine tested the seats at those speeds.

''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 Consumer Reports' Don Mays, a product safety director.

NHTSA Administrator Nicole Nason issued a statement saying: ''We are always interested in making car seats better and safer but not more complicated and difficult for parents. ... We don't want consumers misled into thinking holding a child is better than putting it into a car seat.''

Nine seats failed some or all of the higher-speed tests, Consumer Reports said, while meeting the federal 30 mph standard. Another seat was judged unacceptable because it did not fit well in several cars, the magazine said.

Messages seeking comment were left with the Juvenile Products Manufacturers Association, a trade group.

The magazine tested the type of seat that faces the rear and snaps in and out of a base. It used test dummies weighing 22 or 30 pounds, depending on the seat manufacturers' claims.

In the 35 mph test, seats separated from their bases, rotated too far or would have inflicted grave injuries, Consumer Reports said. At 38 mph, four seats flew out of their bases, it said.

The only seats that passed all the tests were the Baby Trend Flex-Loc and the Graco SnugRide with EPS -- expanded polystyrene foam -- both selling for about $90. Consumer Reports urged parents shopping for seats to buy one of those two, but it also noted that ''any child car seat is better than no seat at all.''

It also said some seats performed better when attached by vehicle safety belts than when attached with the LATCH system. The system, which stands for Lower Anchors and Tethers for Children, includes belts that hook the base of a car seat to metal anchors in the vehicle.

_________

If you are currently using any of the problematic products metioned above, please stop and get a safer seat right away.  Also, I recommend $26 investment for a one-year subscription of CR online - after all, what's $26 compared to a brand new MB or more importantly, an infant's life. 

WASHINGTON, Jan. 18 — Consumer Reports magazine on Thursday retracted an article on infant car seats, published two weeks ago, that said most of them had failed side-impact crash tests.

The tests were supposed to simulate an impact at 38 miles an hour, but actually simulated more than 70 m.p.h., according to the National Highway Traffic Safety Administration, which was so startled by the article that it tried to duplicate the magazine’s results at a government laboratory last weekend. The agency does not have a standard for side impacts but said that at 38 m.p.h., the seats all appeared to do well.

The magazine said it had decided to retract the article after receiving data from the highway traffic agency. The magazine’s tests were performed under contract at an outside laboratory, said a spokesman, Ken Weine, and an internal investigation was under way to determine what went wrong. Consumer Reports did not name the laboratory.

The car seat article was flagged on the cover of the current issue with a headline that said, in red letters, “Safety Alert.” Consumers Union, the nonprofit group that publishes the magazine, distributed the article in advance to newspapers and television stations.

The article said that of a dozen models tested, some fell apart in side-impact tests, and that one model should be recalled by the government because it did not meet the standard for front-impact tests.

The article followed a recent pattern in the magazine of taking on a more journalistic tone, with photos of one baby who was killed in a side crash when he was ejected from his car seat and another who was bruised when his seat separated from its base in a collision.

The seats are rear-facing models for children under a year old, and each has a base that stays in the car. In some tests, the magazine said, the seat part fell off and skittered across the laboratory floor.

The federal government requires that the seats protect babies in front impacts of 30 m.p.h. The highway traffic safety agency said it was trying to develop a side-impact standard. It rates cars under a New Car Assessment Program, which it uses to award “stars” to each model, and those are done at 38 m.p.h. for side impact.

Consumer Reports used seats attached to a sled, rather than to an entire car, in its simulations.

The administrator of the highway safety agency, Nicole R. Nason, said in a statement that “it is absolutely essential for every parent to understand that the safest place in an automobile for an infant is in a car seat.”

Consumer Reports on Thursday affirmed a point made in its original article, that any seat is better than no seat, but told its readers “to suspend judgment on the merits of individual products until the new testing has been completed and the report republished.”

Consumer Reports, an authoritative source on thousands of products, has occasionally made errors. In 1998, it said that some dog foods were deficient in some minerals and should not be used, but after re-testing, it said that they were all nutritionally complete.

In 2006, it published an article that said that six hybrid cars were unlikely to save their owners money, but then said that it had miscalculated depreciation and that two of them would save money.