In a case-control design the feeding in infancy of newly diagnosed 7- to 14-year-old diabetic children (n = 426) was compared with that of age- and sex-matched non-diabetic children (n = 426) randomly selected from the Finnish population registry. All 7- to 14-year-old diabetic children diagnosed from September 1986 to the end of April 1989 from all hospitals which treat diabetic children in Finland were invited to participate in the study. Breast-feeding was initiated in almost all children, but during the birth years of this study population (1972-1982), an increase was observed in the duration of breast-feeding (whether alone or in combination with supplementary feeding) and in the age of introduction of supplementary milk feeding. The risk of Type 1 diabetes was decreased in the children who were totally breast-fed for at least 2 months (odds ratio (OR) 0.64, 95% confidence interval (CI) 0.42-0.98) or 3 months (OR 0.67, 95% CI 0.48-0.95) or exclusively breast-fed for at least 2 months (OR 0.60, 95% CI 0.41-0.89) or 3 months (OR 0.63, 95% CI 0.43-0.93). Those children who were younger than 2 months (OR 1.54, 95% CI 1.08-2.18) or 3 months (OR 1.52, 95% CI 1.11-2.08) at the time when supplementary milk feeding was begun had an increased risk of Type 1 diabetes. These associations remained significant after adjusting for the mother's education. The results suggest that early infant feeding patterns are associated with the risk of Type 1 diabetes developing at the age of 7 to 14 years.