ATLANTA (AP) - New mothers who don't breast-feed and those who smoke after giving birth place their babies at strong risk for Sudden Infant Death Syndrome, government researchers said.

Placing babies on their backs for sleep has long been the focus of the campaign to prevent SIDS, which kills about 3,000 infants each year.

But a study presented at a Centers for Disease Control and Prevention conference in Atlanta suggests that breast-feeding and protecting babies from secondhand smoke may be just as important.

The study of 117 SIDS cases in Louisiana in 1997 and 1998 found 55 percent of the deaths could have been prevented had the mothers breast-fed their children. Studies have shown breast-feeding can help prevent respiratory problems sometimes related to SIDS.

The study also said 27 percent of the deaths could have been prevented if mothers had not smoked after delivery, exposing their children to secondhand smoke.

The link between sleeping position and SIDS was low enough in the study to be deemed statistically insignificant, the authors said.

SIDS - the name coined for unexpected deaths of seemingly healthy babies - kills more infants each year than cancer, heart disease, pneumonia and AIDS combined. Its precise cause has puzzled scientists for decades.

SIDS deaths have dropped by about 40 percent since 1994, when the American Academy of Pediatrics launched a ``Back to Sleep'' campaign to encourage parents not to place infants on their stomachs at bedtime.

But the SIDS death rate appeared to begin leveling off in the late 1990s.

``The `Back to Sleep' campaign may have been effective ... but increasing `Back to Sleep' will not eliminate SIDS,'' said John Painter, a CDC epidemiologist who led the study.

Judy Jacobson, executive vice president of the SIDS Alliance, said placing infants on their backs remains the best way to prevent SIDS.

``There are still large pockets of the population in the United States that do not know about this recommendation - or if they do, have decided not to follow it,'' Jacobson said.

Critics at the conference pointed out that the study excluded the possible link between deaths and the firmness of a child's bedding. The bedding data was thrown out because it was unreliable, Painter said.

Copyright 2001 The Associated Press. All rights reserved.