Background: Type 1 diabetes mellitus is an autoimmune disease in which serum antibodies against islet antigens have been recognized. These antibodies include insulin autoantibodies (IAAs), cytoplasmic islet cell antibodies (ICA) and GAD antibodies. Recently, there has been increasing interest in the use of glutamic acid decarboxylase antibodies (GADA) for the identification of subjects with increased risk of developing type 1 diabetes. GAD antibodies were first discovered in 1982 and is detected persistently after long duration of type 1 diabetes, whereas ICA is transient. However, because the classic immunoprecipitation assays of GAD antibodies is still rather time-consuming, a more simple and reproducible radiolignad binding assay (RBA) is has been widely used recently. The RIA (radioimmunoassay) and IRMA (immunoradiome- tricassay) for GAD antibodies using (125)I-labelled human GAD has been developed, The aim of the present study is to evaluate the usefulness of each methods.Methods: We measured GAD antibodies by RBA with in vitro spathesized recombinani S-methio- nine-labelled GAD65, and protein A-sepharose to separate free from antibody-bound ligand and radioimmunoassay and immunoradiometric assay using 'I-labelled human GAD kit, in addition to measurement of ICAs by standard indirect immunofluorescence technique in 26 patients with type 1 diabetes(male 10, female 16, mean age 14 years) and 10 normal controls(male 5, female 5, mean age 15 years). Results: The overall prevalence of GAD antibodies by RBA and RIA in patients with type 1 diabetes was 38% (10/26), respectively. The prevalence of GAD antibodies by IRMA in patients with type 1 diabetes was 31% (8/26). The frequency of GAD antibodies by RBA,IRMA and RIA increased as the JDF unit of ICA increased. There is a significant correlation between the GAD index (by RBA) and GAD concentration (by RIAand IRMA). Conclusion: These results suggest that GAD antibodies (by RIA or RBA or IRMA) is useful for screening and diagnosis of type 1 diabetes in Korean, but long-term prospective studies on large cohorts of patients is necessary.