In 1987 serum was collected from 1031 non-diabetic schoolchildren in the Southeast area of Sweden with the aim of evaluating islet autoantibody status (ICA, GADA and IA2-ab) in the prediction of diabetes in schoolchildren. The clinical development of Type 1 diabetes in the children was assessed in 1994 and 1997. The combination of ICA, GADA and IA2-ab were found in four subjects whereas six had two and 35 children one of these antibodies. After 10 years, six of the 1031 children had developed clinical diabetes and five of these six children were positive for islet antibodies. Two were positive for all three antibodies, two were positive for ICA and GADA, and one was positive for GADA. Among the individual autoantibodies, ICA showed the highest positive predictive value (29%) whereas the predictive value for the combination of two autoantibodies was highest for GADA and ICA (40%). Thus, GADA and ICA measurements may be a rational approach to detect schoolchildren at risk for developing diabetes.