Objective: To investigate the effect of different antithyroid drug (ATD) regimens on relapse rates of Graves' disease, and to look for predictors of relapse. Design and Methods: In a prospective two-way factorial randomized clinical trial, 218 patients with Graves' disease were assigned to ATD combined with L-thyroxine (L-T 4 ) or ATD alone for 12 months. After discontinuation of antithyroid therapy, each group was stratified to either 12 months further treatment with L-T 4 or no treatment. Clinical and biochemical assessments were carried out before treatment, after 3 and 6 weeks, and every third month for 12 months. If the patients lacked symptoms of relapse, laboratory tests were performed every third month for the second year, and thereafter annually. Results: The proportion of all patients with relapse was 47.7% 2 years after withdrawal of ATD. There was no difference in relapse rates between the treatment groups (P ¼ 0:217; log-rank test). Smokers had a higher relapse rate than non-smokers (58.4% vs 38.8%, P ¼ 0:009). Patients who were thyrotropin-receptor antibody (TRAb) positive after 12 months of antithyroid therapy had a higher relapse rate than those who were negative (72.5% vs 36.8%, P , 0:0001). Similarly, relapse was more frequent (55.5%) in patients having large goiter compared with subjects with small goiter (36.3%, P ¼ 0:0007). Conclusions: Relapse rates of Graves' disease were independent of ATD regimen whether followed by L-T 4 therapy or not. Smoking, large goiter and presence of TRAb at the end of ATD therapy were strong predictors of relapse.