The relationship of development of diabetes from birth up to 15 years of age to the type of feeding in infancy, childhood infections and vaccination was studied in 55 patients attending Endocrinology Clinics of the Ethio-Swedish Children's and Tikur Anbessa Hospitals over a period of two years (January 1990 to December 1991). Seventy-four unaffected siblings and 107 unrelated controls were interviewed for comparison. No significant difference was found in relation to type of feeding up to the ages of three, six and 12 months or older between patients and unaffected siblings. Histories of measles, chicken pox and whooping cough were equally distributed between the two groups. However, introduction of bottle-feeding was significantly more frequent among unrelated controls at three months of age (9/39 diabetics versus 41/83 controls) and six months of age (26/39 diabetics vs 72/83 controls) chi 2 = 6.6, (p < 0.01) and chi 2 = 5.6 (p < 0.05) respectively. The odds ratios between diabetics and unrelated controls for introduction of bottle-feeding at three months and six months of age were 0.32 (confidence intervals 0.14-0.74) and 0.31 (confidence intervals 0.13-0.77) respectively. The use of cow's milk and other formulas in bottle-feeding showed a significant negative association with the development of diabetes chi 2 = 5.8 (p < 0.025), chi 2 = 3.8 (p < 0.05) respectively. A history of vaccination against tuberculosis, measles, diphtheria, pertussis, tetanus (DPT) and polio was significantly more common among unrelated controls than diabetics, chi 2 from 4.6 to 11.4 (p < 0.05 to p < 0.001).(ABSTRACT TRUNCATED AT 250 WORDS)