Glutamic acid decarboxylase antibodies (GADAbs) are being increasingly used in clinical and research programs for the prediction and classification of insulin-dependent diabetes mellitus (IDDM). A number of different assay formats for the measurement of GADAbs have been reported, but the degree of concordance between assays is unknown. In this study, GADAbs were measured on 16 coded sera in 34 assays to examine concordance between GADAb assays and establish the feasibility of an international GADAb, standard of measurement unit. The 16 lyophilized coded samples consisted of sera from healthy control subjects (n = 2), IDDM patients (n = 3), a patient with polyendocrine autoimmunity (n = 1), and duplicate dilutions of plasmapheresis serum from a patient with stiff-man syndrome (SMS). A high level of concordance was found in the ranking of GADAb levels (P = 0.99, Friedman's test) in the samples. Thirteen (38%) assays could reproducibly distinguish dilutions of SMS serum and detect GADAbs in all IDDM and polyendocrine autoimmunity sera tested. Although assessed on only four samples, disease specificity was 100% in 29 assays. The majority of assays that immunoprecipitated radiolabeled GAD gave high results for sensitivity and specificity. Enzyme-linked immunosorbent assays and assays using immunofluorescence were generally less sensitive. Several assays, in particular those measuring GAD enzymatic activity immunoprecipitated in fluid phase from rat brain homogenate, showed a prozone-like phenomenon in the SMS dilution curve. Interpolation of results from a standard curve into workshop units resulted in relatively low scatter in samples with lower levels of GADAbs. Hence, the use of an international reference serum to enable comparison of results between laboratories appears feasible. A second serum exchange is currently in progress to explore differences in sensitivity and specificity using larger numbers of disease and control sera.