The evidence on the role of environmental factors in the development of type 1 diabetes is conflicting. Reducing potential bias and the variety of exposures we investigated the association between type 1 diabetes risk and nutritional and other environmental exposures in preschool children. This nationwide case-control study included 760 cases newly diagnosed with type 1 diabetes under five years of age during 1992-1995. 630 controls of the same age were selected from the case families' acquaintance. Information on infant diet, foetal, perinatal and socio-economic factors, and family history of diabetes was obtained by a parent-administered questionnaire. Data were analysed by multiple unconditional logistic regression. Duration of breastfeeding and age at introduction of bottle-feeding were inversely associated with type 1 diabetes risk according to a dose-response relationship (trend test p<0.05). Adjusted odd ratios (95%-CI) for a short breastfeeding period and an early introduction of formula feeding (<5 vs. > or =5 months) were 1.31 (1.01-1.69) and 1.34 (1.03-1.74), respectively. Familial type 1 diabetes was found more frequently among diabetic than among control children. Higher social status, late introduction of solid food (> or =5 month), and higher current cow's milk consumption (> or =200 ml/d) were associated with a reduced diabetes risk. A considerable proportion of the diabetic risk among preschool children was explained by modifiable exposures. Our findings indicate that infant feeding is causally associated with type 1 diabetes risk and that a considerable part of new type 1 diabetic cases is potentially preventable.