This study has investigated the genetic basis of the heterogeneous autoimmune response to glutamic acid decarboxylase (GAD) in 179 Australian patients with IDDM. Antibodies to GAD have been correlated with HLA-DQB1 alleles and genotypes, as determined by sequence-specific oligonucleotide hybridizations after polymerase chain reaction was applied to exon 2 of the DQβ1 gene. HLA-DQ2 was significantly increased (p τ 0.01) in IDDM patients with antibodies to GAD. Antibodies to GAD were detected in 64% of 72 DQ2, 8 patients, in 55% of 29 DQ2, 2 or DQ8, 8 patients and in 41% of 78 patients with other HLA-DQB1 genotypes. HLA-DQ genotype association with autoimmunity to GAD was statistically significant (p = 0.02) and reflected early formation of antibodies to GAD, rather than an HLA association with persistence of antibodies to GAD, since the genotype effect was more evident (p = 0.02) in those with more recent onset (0-5 years) of IDDM. Also, the HLA-DQ genotype effect was more evident in patients with IDDM onset after the age of 14 years (p = 0.003). Multivariate analysis showed that HLA-DQB1 genotypes had a more significant impact on antibodies to GAD than either duration or age of onset of IDDM. In patients with IDDM in childhood, only a minority had low-risk HLA-DQB1 genotypes (37%) when compared with those with onset in adulthood (62%) (p = 0.005). The possibility of preclinical screening for IDDM among children is supported by our finding of a genetic basis for autoimmune response to GAD, noting that the majority of prediabetic children have high-risk HLA-DQB1 genotypes and these correlate with antibodies to GAD.