The Sardinian population has an extremely high incidence of IDDM (30.2 of 100.000 in the age group of 0-14 years). This study reports the molecular characterization of HLA class II genes in 120 IDDM sporadic patients and 89 healthy subjects of Sardinian origin. Compared with other Caucasians, both Sardinian patients and controls had an unusual distribution of haplotypes and genotypes. In particular, there was a high gene frequency of the DRB1*0301, DQA1*0501, DQB1*0201 susceptibility haplotype both in patients (0.58) and controls (0.23) while a reduction of the DRB1*1501, DQA1*0102, DQB1*0602 protective haplotype (0.03) was observed in the healthy population. This distribution may partially explain the high incidence of IDDM reported in Sardinia. The analysis of the DQ beta 57 and DQ alpha 52 residues showed that the absence of Asp 57 and the presence of Arg 52 were associated with IDDM in a dose-response manner. On the other hand, we found that (a) a very similar distribution of these residues was found when comparing Sardinians with another healthy Caucasian population from the same latitude but with a lower rate of IDDM incidence; (b) several genotypes encoding the identical DQ alpha 52/DQ beta 57 phenotype carried very different relative risks; and (c) the DRB1*0403, DQA1*0301, DQB1*0304 haplotype (DQ beta 57 Asp-neg and DQ alpha 52 Arg-pos) was found in 40% of the DR4-positive controls but not in patients (p = 0.00034), while the DRB1*0405, DQA1*0301, and DQB1*0302 haplotype carrying the same residues at the same positions was found in 70% of the DR4-positive patients and in only one control (p = 0.00003). These findings suggest that IDDM susceptibility cannot be completely explained by the model in which only DQ alpha 52 and DQ beta 57 residues are taken into account.