Blood was drawn from 74 children, 3-16 years old, at diagnosis of Type 1 (insulin-dependent) diabetes and before the first insulin injection. Insulin autoantibodies were detected with a polyethylen-glycol-method in 27/74 (36.4%) and with an immuno-electrophoretic method in 6/74 (8.1%). Islet cell cytoplasmic antibodies detected by indirect immuno-fluorescence were found in 49/74 patients (66.2%), who included as many as 23 of the 27 patients with insulin autoantibodies determined with the polyethylen-glycol-method (p less than 0.01). The proportion of insulin autoantibody-positive patients who developed insulin antibodies during the first 9 months of insulin treatment was not significantly greater (51.8%) than that of insulin autoantibody-negative patients (44.6%), but patients with both islet cell antibodies and insulin autoantibodies at diagnosis produced more insulin antibodies during the first 9 months (p less than 0.05). There was no difference in fasting or meal stimulated serum C-peptide after 3, 9 or 18 months as related to occurrence of insulin autoantibodies and/or islet cell antibodies. The correlation between insulin autoantibodies and islet cell antibodies indicates that both types of autoantibodies reflect the same immunological process, although the lack of correlation to C-peptide may indicate that they play a minor causal role. In addition, the results show that patients with an active autoimmune process evidently tend to produce more insulin antibodies during the first months of insulin treatment, but the islet cell antibodies and insulin autoantibodies-positive patients had at least as good residual B-cell function as patients without autoantibodies at diagnosis.(ABSTRACT TRUNCATED AT 250 WORDS)