To establish an improved radioligand assay of insulin autoantibody (IAA) and to investigate the diagnostic role of IAA in type 1 diabetes (T1DM).125I-insulin antigens were mixed with sera samples in the Eppendorf tubes. The immunocomplexes were precipitated with protein A-agarose, then washed with TBST buffer and counted the cpm value using liquid scintillation counter. The assay results were expressed by IAA index and compared with those of an international standardized laboratory and a domestic anti-insulin antibody kit. Sera of 32 recent-onset type 1 diabetic patients and 120 healthy controls were screened for IAA and the consistency, sensitivity, specificity and the diagnostic value of IAA were evaluated.(1) The intra-assay coefficient of variation (CV) was 5.8% - 8.3%, and the inter-assay CV 7.0% - 11.0%. The IAA indices of 41 samples tested in our laboratory and in an international standardized laboratory were strongly positively related (r = 0. 783, P < 0. 01) (100.0% concordance). Arrayed according to their antibody titers, the IAA indices of 82 samples measured in our laboratory and in a domestic anti-insulin antibody kit were also strongly positively related (r = 0.982, P < 0.0001) (89.0% concordance, Kappa value 0.813), but 9 samples were positive with low titer (index 0.05 - 1.00) by the radioligand assay or negative by the kit. 2) The IAA index cut-off point was determined according to the upper limit of 99.5% percentile of 120 healthy controls. The index of 0.051 or higher was defined as positive. The IAA positivity prevalence was 18.8% (6/32) in type I diabetes and 0.8% (1/120) in the healthy controls (P < 0.001). The positivity frequency of IAA was 28.6% (4/14) in patients younger than 15 years of age and the duration less than half a year, while 11.1% (2/ 18) in those older than 15.The improved radioligand assay for IAA has a high sensitivity and specificity and can be used in clinical practice, especially for the low-titer positive subjects.