Prediction of Type 1 diabetes for study of preventive therapies requires screening the general population, where 85% of new cases occur. Even with HLA-based prescreening, nearly 20% of all children will need multiple serum autoantibody testings. High-throughput, economical, and accurate methods are therefore essential. We have developed such a radiobinding method, using 96-well microtiter plates and a novel immune complex capture method via membrane-bound Protein A. Each microtiter plate contained a standard negative control serum, and low-, medium-, and high-level positive control sera. All sera were evaluated in triplicate. This readily allowed quality control criteria both for triplicates of individual sera and for each 96-well plate. Inter-assay coefficients of variation (CVs) were all </=16%, while intra-assay CVs were all </=10%. The assay was found to be sensitive (to detect autoantibodies in patients) and specific (low reactivity in thousands of healthy volunteers). The format worked well using diverse antigens such as 35S-met-GAD65, 35S-met-ICA512/IA2, 35S-met-Phogrin, and 125I-insulin, and could be used for simultaneous screening of reactivity to both GAD65 and ICA512/IA2 in the same well. Diagnostic accuracy compared favorably with microcentrifuge tube-based Protein A-agarose GAD65 and IA2 autoantibody radiobinding assays and with acid-charcoal-polyethylene glycol (PEG) based competitive insulin autoantibody assays. In the case of 125I-insulin, comparing signal in the absence versus presence of cold insulin competitor was not necessary. Total serum volumes required were only 6 microl for GAD and ICA512, and only 15 microl for IAA. The method costs less than all other commonly used formats, and should be useful for population screening.