The majority (about 90%) of children developing Type 1 (insulin-dependent) diabetes mellitus do not have a first-degree relative with the disease. Nearly all (389/405, 96%) children (0-14 years) in Sweden, who developed diabetes during one year, were therefore studied to compare islet cell, thyroid peroxidase, thyroglobulin, and gastric H+, K+-ATPase antibodies with 321 age, sex, and geographically matched, but non-related, control children. Islet cell (cytoplasmic) antibodies were found in 81% (316/389) of the patients and in 3% (9/321) of the control children (p less than 0.001). The median islet cell antibody levels were 70 (range 3-8200) Juvenile Diabetes Foundation (JDF) Units in the islet cell antibody positive patients, and 27 (range 17-1200) JDF Units in the control children (NS). Autoantibodies against thyroid peroxidase (8%), thyroglobulin (6%), and gastric H+, K+-ATPase (3%) were all increased in the patients compared with the control children, being 2% (p less than 0.001), 2% (p less than 0.01), and 0.3% (p less than 0.01), respectively. During an observation time of 20-34 months, two of the nine islet cell antibody positive control children developed Type 1 diabetes, after 8 and 25 months respectively, while the others remained healthy and became islet cell antibody negative. None of the islet cell antibody negative control children developed diabetes during the same time of observation. This first investigation of an unselected population of diabetic children and matched control children shows: that islet cell antibodies are strongly associated with newly diagnosed childhood diabetes, that other autoantibodies are more frequent among diabetic children than control children, and that the frequency of islet cell antibodies in the background population of children is higher than previously documented, and could also be transient, underlining that factors additional to islet cell antibodies are necessary for the later development of Type 1 diabetes.