A study found that the longer women nursed, the lower their risks of developing diabetes.
The findings are far from conclusive, but the researchers say breast-feeding may change mothers' metabolism in ways that make the possible connection plausible.
These metabolic changes may help keep blood sugar levels stable and make the body more sensitive to the blood sugar-regulating hormone insulin, said Dr. Alison Stuebe, the study's lead author and a researcher at Boston's Brigham and Women's Hospital.
That theory is partly based on evidence in rats and humans showing that breast-feeding mothers had lower blood-sugar levels than those who did not breast-feed.
The new study, published in Wednesday's Journal of the American Medical Association, involved 157,000 nurses who participated in two long-running health studies. They filled out periodic health questionnaires and were followed for at least 12 years. During the study, 6,277 participants developed type 2 diabetes.
Women who breast-fed for at least one year were about 15 percent less likely to develop type 2 diabetes than those who never breast-fed. For each additional year of breast-feeding, there was an additional 15 percent decreased risk.
However, both breast-feeders and bottle-feeders studied faced very low absolute risks of developing the disease.
In the first study, which began in 1976, 6.3 percent of women who breast-fed less than one year or not at all developed diabetes, compared with 5.5 percent of women who breast-fed for more than a year. In the second study, which began in 1989, the rates were 1.9 percent and 1.1 percent respectively.
"If it does have an effect, it's very small," said Dr. Lisa Schwartz of Dartmouth Medical School, co-director of a research group that studies how medical information is sometimes hyped. She was not involved in the breast-feeding study.
With diabetes the nation's sixth-leading cause of death and 82 million U.S. women of childbearing age, even a small risk reduction could have a big effect, Stuebe said.
Continuous breast-feeding for at least one year appeared to be slightly better than breast-feeding each child for shorter durations, but the differences were minimal, Stuebe said.
Schwartz said the results may reflect the healthy lifestyles of women who breast-feed rather than breast-feeding itself. But the researchers said that taking habits such as exercise, diet and smoking into account did not change the results.
Dr. Ruth Lawrence of the University of Rochester in New York, author of a medical textbook on breast-feeding, called the results compelling.
She noted that previous research has suggested that breast-feeding might reduce women's risk of breast and ovarian cancer and osteoporosis. If diabetes could be added to that list, the effect would be substantial, Lawrence said.
Breast-feeding has numerous health benefits for babies, too, so encouraging mothers to nurse "is kind of a win-win from a public health standpoint," Stuebe said.