Reliable genetic and immunological markers are important in the prediction of insulin-dependent diabetes mellitus (IDDM). Since glutamic acid decarboxylase (GAD) is a candidate primary autoantigen, we examined the possible linkage between IDDM and the genes encoding GAD₆₅ (GAD2, 10p11–12) and GAD₆₇ (GAD1, 2q31) in 58 Danish IDDM affected sib pairs. The allelic inheritance of 10 polymorphic dinucleotide repeat sequences spanning the chromosomal regions of the two GAD genes, were examined by affected sib pair analysis (ASP). In addition a restriction fragment length polymorphism (RFLP) was identified in the gene encoding GAD₆₅ using the restriction enzyme PvuII. The GAD gene markers were analyzed in relation to the presence of specific HLA types and GAD autoantibodies. No evidence of linkage was found between IDDM and either of the genes encoding GAD. This was also the case when subgroups carrying specific HLA susceptibility alleles were analyzed. Nor did we observe any association between these GAD genetic markers and the presence of GAD autoantibodies. Considering the high prevalence of GAD autoantibodies in IDDM, a putative genetic association between GAD and IDDM would be expected to affect most diabetic individuals. Therefore, our data indicate that the association between GAD and IDDM is not genetically determined, and that microsatellites used in this study do not contribute to the prediction of IDDM.