In the Asian populations, it is not uncommon for adult patients with NIDDM to eventually lose beta-cell function and develop IDDM. Accepting that IDDM is an autoimmune disease, which occurs on a genetic background, it could by hypothesized that by measuring autoantibody prevalence and HLA DQ gene polymorphism, known important prediagnostic markers of IDDM, the prevalence of adult-onset IDDM in patients with previously undiagnosed NIDDM patients could be estimated. To do this, anti-GAD prevalence and HLA DQ A1 and DQ B1 polymorphisms after PCR amplification of genomic DNA were analyzed in 121 newly diagnosed diabetic patients of Yonchon cohort and compared to the results with those of 100 matched health control subjects. We also compared the results with those of other populations to assess the difference of genotype distribution. The overall prevalence of anti-GAD antibodies was 1.7% (2 of 121) in patients with previously undiagnosed NIDDM, whereas 1 of 100 controls had positive antibodies. Among those who were positive, their titer of antibodies to GAD were not high. No statistically significant differences in the distribution of either mean levels of anti-GAD or DQA1 and DQB1 alleles were found comparing NIDDM patients to controls. Interestingly, the frequency of DQB1*non-Asp-57 and DQA1*Arg-52 alleles in the Korean adult control population was similar to that of US Caucasians (DQB1*non-Asp-57: 0.431 vs. 0.475; DQA1*Arg-52: 0.492 vs. 0.463). The low prevalence of anti-GAD antibodies and HLA-DQA1 and DQB1 susceptibility alleles among recent-onset NIDDM patients, not different compared to controls suggests that diabetes in Korean adults is unlikely to have an autoimmune component to its pathogenesis.