In order to test the hypothesis that breast-feeding has a protective effect on the development of insulin-dependent diabetes (IDDM) during childhood we retrospectively studied 297 diabetic children age 15 years or less diagnosed 1974-88 at the 5 pediatric departments in the South-East region of Sweden. They were compared to 792 non-diabetic controls of the same age (year of birth), sex and geographical location. Files from the child welfare clinics were reviewed and the families answered questionnaires. Ten percent of the diabetic children and 11% of the control children had never been breast-fed. In the total material no obvious difference was seen regarding duration of breast-feeding between the diabetic children and their healthy controls. Only in the oldest maternal age group (> 35 years of age) control mothers tended to have a longer duration of breast-feeding (P < 0.01). There was also a trend that children with diagnosis of IDDM during winter had been breast-fed for a shorter period (4.1 +/- 3.4 months) than their controls (5.0 +/- 3.5 months, P < 0.09). We conclude, that breast-feeding has very little effect on preventing IDDM in children. If it has any effect, it might be to decrease the risk in certain subgroups.