To elucidate whether autoantibodies can be used to predict the intensity of autoimmune beta-cell destruction, we determined both C-peptide and autoantibodies (islet cell antibodies (ICA), insulin autoantibodies (IAA), islet cell surface antibodies (ICSA) and antibodies to glutamic acid decarboxylase (GADA)). In 89 diabetic children and adolescents at diagnosis at the age of 1.2-16.6 years (mean +/- S.D., 9.0 +/- 4.5). Only 12/89 (14%) had no autoantibodies at diagnosis, while 2 patients (2%) had all 4 autoantibodies. There was a positive correlation between GADA and ICA (P < 0.01). At diagnosis 70% of the patients had GADA, most common in patients above the age of 8 years at diagnosis (P < 0.001), and with higher GAD-index in girls (P < 0.05). ICA was detected in 63%, most common in the older age groups (P = 0.04). ICSA seen in 22% of the patients as well as IAA (detected in 32%) were most common < 8 years of age (P = 0.06, P = 0.08, respectively). Children with autoantibodies had similar C-peptide levels through the follow up period as children of the same sex and age without antibodies, except for patients with ICSA alone or in combination with other autoantibodies who tended to have higher C-peptide levels. We conclude that not even combinations of autoantibodies can be used to predict beta-cell destruction in IDDM patients.