Drug-based therapy is usually the initial treatment for Graves' disease (GD) hyperthyroidism in children, but there is some debate about treatment duration. Our objective was to assess the effect of long-term carbimazole therapy on GD remission in children and its determinants. This was an observational prospective multicenter follow-up cohort study. Participants included 154 children newly diagnosed with GD between 1997 and 2002. The intention was to treat patients with three consecutive courses of carbimazole, each lasting 2 yr. Definitive treatment was performed in cases of poor compliance with antithyroid drug (ATD) treatment, thyrotoxicosis relapse, or major adverse effects of ATD treatment. The main outcome measure was remission for at least 18 months after the completion of each course of ATD treatment. The median duration of follow-up was 10.4 (9.0-12.1) yr. Overall estimated remission rates (95% confidence interval) 18 months after the withdrawal of ATD treatment increased with time and were 20 (13-26), 37 (29-45), 45 (35-54), and 49 (40-57)% after 4, 6, 8, and 10 yr follow-up, respectively. A multivariate competing risk model revealed an independent positive effect of less severe forms of hyperthyroidism at diagnosis [subhazard ratio of 1 for patients with free T(4) <35 pmol/liter vs. 0.4 (0.20-0.80) for free T(4) ≥ 35 pmol/liter; P = 0.01] and of the presence of other autoimmune conditions [subhazard ratio of 2.23 (1.19-4.18); P = 0.01] on remission rate after medical treatment. About half the patients achieved remission after carbimazole discontinuation, and there seems to be a plateau in the incidence of remission achieved after 8-10 yr ATD therapy.