The WHO DiaMond Molecular Epidemiology Sub-Project is testing the hypothesis that the geographic differences in IDDM incidence reflect population variation in the frequency of IDDM susceptibility genes (i.e., DQA1 and DQB1 alleles with sequences coding for arginine (R) in position 52 of the DQ α-chain, and an amino acid other than aspartic acid (ND) in position 57 of the DQ β-chain, respectively) using a standardized case-control design. Data from twelve populations which have completed (or have nearly completed) recruitment and HLA molecular analyses are presented. There was an approximate 2-fold increase in the frequencies of DQA1*0301, DQB1*0201 and DQB1*0302 among IDDM cases compared to non-diabetic controls in most populations. Interestingly, DQA1*0301 was more common in low versus moderate-high incidence countries. DQB1*0201 and DQB1*0302 were more prevalent in the moderate-high incidence areas. DQA1*R and DQB1*ND were both consistent markers of IDDM risk, with stronger associations in moderate-high versus low incidence areas. In general, individuals homozygous for both DQA1*R and DQB1*ND had the highest genotype-specific IDDM incidence rates, which approximated risk estimates for first degree relatives in several countries. These data revealed considerable variation in the frequencies of DQB1 and DQA1 alleles across countries, which likely contribute to the global patterns of IDDM incidence.