Antithyroid drugs are commonly used as first-line treatment for Graves' disease, but the optimum regimen for inducing remission remains unclear. We gave the block-replace regimen of carbimazole plus thyroxine to 100 patients for 6 or 12 months, to determine whether prolonged treatment is associated with fewer relapses. The remission rate one year after cessation of treatment was 59% with the 6 month course and 65% with 12 months; this was not significantly different. We also analysed HLA markers identified by restriction fragment length polymorphisms and could not confirm the recently reported associations of outcome with HLA-DR4 or with an HLA-DQA2 allele. These results show that six months treatment with a block-replace regimen of antithyroid drugs is probably sufficient, in the UK, to achieve maximum remission of Graves' disease and that there are no HLA markers which clearly predict outcome.