Recently, 65-kDa glutamic acid decarboxylase (GAD 65) antibodies (GADA) have been introduced as autoimmune markers in blood to confirm the diagnosis of insulin-dependent diabetes mellitus (IDDM). In this study, to evaluate two new assays that use 125I-labeled GAD 65, we assayed samples from 100 children with recent onset of diabetes and 100 control children, the results were compared with those of a [35S]GADA assay and with results for islet cell antibodies (ICA), the conventional autoimmune marker. Receiver operating characteristic (ROC) curve analysis showed one of the new assays (from RSR) to be more sensitive (P = 0.01) than the comparison ([35S]GADA) assay, whereas the second new assay (from Elias) was less sensitive (P < 0.001). The GADA frequency at 97.5% specificity was greatest in the comparison assay63 of 100 vs 41 of 100 (P < 0.01) and 53 of 100 (P = 0.16) in the RSR and Elias assays, respectively. Almost all GADA-positive patients had ICA, but one-third of the ICA-positive patients was GADA-negative. Accordingly, adding GADA analysis results to ICA testing increased the frequency of detection of autoimmune markers only slightly (from 81% to 85%). In conclusion, at 97.5% specificity the [35S]GADA assay seemed to be more efficient than the 125I assays, although the difference was significant only for the Elias 125I assay. Antigen-specific antibodies other than GADA may explain the difference in GADA and ICA frequencies.