Type I (insulin-dependent) diabetes mellitus is an autoimmune disease characterised by the presence of circulating islet autoantibodies and by T-lymphocyte infiltration of the islets of Langerhans with consequent destruction of pancreatic beta cells [1]. A permissive genetic background is required for the development of the islet autoimmune process. Type I diabetes is a complex genetic disease and a long series of genes has been found to be linked with disease Diabetologia (2000) Methods. Type I diabetic (n = 95) and healthy (n = 98) Italian subjects were typed for exon 5 of MIC-A and for HLA-DRB1, HLA-DQA1, HLA-DQB1 and TNFA alleles. All subjects were also typed for the presence of HLA-B8 or HLA-B15.Results. The frequency of MIC-A5 was increased in diabetic subjects (53 % vs 15 %) (OR = 6.1) (corrected p, p c < 0.0005). Among HLA class II haplotypes, both HLA-DRB1*03-DQA1*0501-DQB1*0201 (DR3-DQ2) and DRB1*04-DQA1*0301-DQB1*0302 (DR4-DQ8) (ªat-risk class II haplotypesº) were positively associated with diabetes (OR = 6.7 and 6.0, respectively) (p c < 0.003). Also HLA-B8 was more frequent among Type I diabetic subjects than among healthy control subjects (OR = 2.8, p = 0.01). None of the TNFA alleles were statistically significantly associated with Type I diabetes. The MIC-A5 exon was negatively associated with age at clinical onset of diabetes (p = 0.012). Thus, 68 % diabetic subjects younger than 25 years and 29 % older than 25 years were carrying this allele. Both MIC-A5 and the at-risk class II haplotypes were independently associated with Type I diabetes and the combined association of the two markers had the highest relative risk (OR = 172). In subjects younger than 25 years, the OR of MIC-A5 was as high as 21.7 and was more than twofold that of at-risk class II haplotypes (OR = 9.5). The MIC-A5 exon was not in linkage disequilibrium with any of the HLA-class I, class II or TNFA alleles studied. Conclusions/interpretation. The MIC-A gene polymorphism is associated with genetic risk for Type I diabetes and the combination of MIC-A5 and at-risk class II haplotypes is now to be seen as the strongest genetic marker for this disease. [Diabetologia (2000) 43: 507±514]