OBJECTIVE: To analyze the HLA-DQ (human leukocyte antigen) genetic association with insulin-dependent diabetes mellitus (IDDM) patients of the Northeast Italian population. RESEARCH DESIGN AND METHODS: Fifty-one IDDM patients and 52 healthy control subjects were molecularly typed for DQB1 and DQA1 loci by using allele-specific oligonucleotide probes and polymerase chain reaction amplified genomic DNA. DNA enzyme immunoassay was used to assess allele specificities. RESULTS: IDDM status strongly correlated with DQB1 alleles carrying a non-aspartic acid (non-Asp) residue in position 57 of DQ beta-chain and DQA1 alleles with an arginine (Arg) residue in position 52 of DQ alpha-chain. Individuals with two DQB1 (non-Asp) alleles and two DQA1(Arg) alleles had the highest relative risk for disease: they constituted approximately 40% of IDDM patients compared with 0% of control subjects. Heterozygosis at either residue 57 of DQB1 or residue 52 of DQA1 was sufficient to abrogate statistical significance for disease association, although 47% of IDDM patients were included in these two groups compared with 21% of normal control subjects. On the other hand, the presence of two DQB1 alleles with Asp in position 57 was sufficient to confer resistance to disease irrespective of the DQA1 genotype. CONCLUSIONS: The results demonstrate that the complete HLA-DQ genotype, more than a single DQB1 or DQA1 locus, should be determined to estimate the highest risk for disease. Screening a population for preventive purposes and/or early signs of IDDM should then take advantage of this result, and "susceptible homozygous" individuals should be followed very closely and considered the first group of choice for possible new therapeutic trials.