Association of the major histocompatibility complex (MHC) class II-encoded HLA-DRB1-DQA1-DQB1 haplotype with Graves' disease (GD) has been known for several years. Recent evidence from other autoimmune diseases has suggested that the HLA class I encoded HLA-B/-C molecules could be conferring HLA-DRB1-DQA1-DQB1 independent effects on disease. The aim of this study was to determine the effect of HLA-B and HLA-C in GD in a white ethnic group of 806 patients with GD and 487 control subjects from the UK. Of the five loci (HLA-B, -C, -DRB1, -DQA1, -DQB1), HLA-C demonstrated the strongest association (P = 1.20 x 10(-20)) with HLA-C*07 predisposing [OR = 1.63, 95% CI (1.23-2.17)] and both HLA-C*03 [OR = 0.54, 95% CI (0.38-0.77)], HLA-C*16 [OR = 0.36, 95% CI (0.21-0.61)] protective. The other loci were then tested for HLA-C-independent associations. HLA-B was found to be associated independently of HLA-C (P = 1.54 x 10(-6)) with the other three loci, HLA-DRB1, HLA-DQB1 and HLA-DQA1, also improving the model but with less confidence (P > 10(-5)). This study has for the first time provided evidence of a primary association of HLA-C, and to a lesser extent HLA-B, with GD. Class II loci could still have effects on GD, but they appear smaller than the HLA-C association. A full investigation of the MHC region, including all class I and II loci is now required. Our results point to a primary role for class I-mediated responses in GD, a condition classically assumed to be a straightforward HLA-class II-restricted autoantibody response to the thyroid stimulating hormone receptor.