Background: A sensitive ELISA for measurement of IA-2 autoantibodies has been developed and assessed. Also, a combination ELISA for detection of both GAD(65) autoantibodies and IA-2 autoantibodies is described. Methods: The IA-2 autoantibody assay is based on the ability of IA-2 autoantibodies to form a bridge between IA-2 intracellular fragment coated onto ELISA plate wells and liquid-phase IA-2 labelled with biotin. The combination ELISA uses plates coated with both IA-2 and GAD65 and a mixture of IA-2-biotin and GAD(65)-biotin. Assay sensitivity was assessed using the WHO reference (NIBSC 97/550) for islet cell antibodies. IA-2 autoantibody measurements by ELISA were compared with measurements in immunoprecipitation assays (IPAs) based on I-125 or S-35 labelled IA-2. Combination ELISA results were compared with results obtained for individual autoantibodies. Results: As little as 15 units/mL of NIBSC 97/550 was detectable in the IA-2 autoantibody ELISA compared to 125 units/mL by I-125-IA-2 IPA. 110/216(51%) sera from patients with type 1 DM were positive in the IA-2 autoantibody ELISA while 97/216 (45%) and 91/216 (42%) were positive in the I-125-IA2 and S-35-IA-2 IPAs, respectively. The IA-2 autoantibody ELISA showed 100% specificity for type 1 DM. The combination ELISA was able to detect GAD(65) and/or IA-2 autoantibodies in 183/216 (85%) diabetic sera and 183/216 (85%) were also found positive for autoantibodies to IA-2 and/or to GAD65 in the assays for individual antibodies. Autoantibody measurements in the individual autoantibody assays and in the combination ELISA showed good agreement by Pearson correlation (r=0.92, n=216, p < 0.001) and by Bland and Altman analysis. Conclusions: Sensitive and specific ELISAs for measurement of autoantibodies to IA-2 and to a combination of IA-2 and GAD(65) have been developed. These assays are suitable for screening large numbers of samples in diabetes prediction and prevention trials. (c) 2005 Elsevier B.V. All rights reserved.