A study of antithyroid drug (ATD) therapy with a mean follow-up period of 10 years (range 2-25) in 434 patients with Graves' disease has been made by linking hospital records with those of a central follow-up register. The majority (89%) were treated with carbimazole and 87% received combined therapy with triiodothyronine (T3) (73%) or thyroxine (T4) (14%). Sixty-one per cent were assessed for T3 suppression tests on completion of treatment, of whom 61% (95% CL, 55-67%) suppressed. The overall 5-year cumulative proportion developing recurrent hyperthyroidism was 54-62% with rates of 26-44% in suppressed patients and 65-79% in those not suppressed. In unsuppressed patients, most (72%) of the recurrences occurred within 1 year with only an additional 10% predicted up to 10 years. In suppressed patients 30% of recurrences occurred in the first year, 60% between 1 and 5 years and a further 10% between 5 and 10 years. Suppression with T3 is probably the best and cheapest predictor of outcome but has an accuracy of only 70% for both positive and negative tests which limits its usefulness in planning long-term follow-up and surveillance. A standard format should be adopted for the analysis and reporting of follow-up studies, based on actuarial methods of estimating the cumulative proportion with recurrences or other events, to facilitate comparisons between different centres.