Atypical antipsychotic adjunctive therapy to lithium or valproate is effective in treating acute mania. Although continuation of atypical antipsychotic adjunctive therapy after mania remission reduces relapse of mood episodes, the optimal duration is unknown. As many atypical antipsychotics cause weight gain and metabolic syndrome, they should not be continued unless the benefits outweigh the risks. This 52-week double-blind placebo-controlled trial recruited patients with bipolar I disorder (n=159) who recently remitted from a manic episode during treatment with risperidone or olanzapine adjunctive therapy to lithium or valproate. Patients were randomized to one of three conditions: discontinuation of risperidone or olanzapine and substitution with placebo at (i) entry ('0-weeks' group) or (ii) at 24 weeks after entry ('24-weeks' group) or (iii) continuation of risperidone or olanzapine for the full duration of the study ('52-weeks' group). The primary outcome measure was time to relapse of any mood episode. Compared with the 0-weeks group, the time to any mood episode was significantly longer in the 24-weeks group (hazard ratio (HR) 0.53; 95% confidence interval (CI): 0.33, 0.86) and nearly so in the 52-weeks group (HR: 0.63; 95% CI: 0.39, 1.02). The relapse rate was similar in the 52-weeks group compared with the 24-weeks group (HR: 1.18; 95% CI: 0.71, 1.99); however, sub-group analysis showed discordant results between the two antipsychotics (HR: 0.48, 95% CI: 0.17; 1.32 olanzapine patients; HR: 1.85, 95% CI: 1.00, 3.41 risperidone patients). Average weight gain was 3.2 kg in the 52-weeks group compared with a weight loss of 0.2 kg in the 0-weeks and 0.1 kg in the 24-weeks groups. These findings suggest that risperidone or olanzapine adjunctive therapy for 24 weeks is beneficial but continuation of risperidone beyond this period does not reduce the risk of relapse. Whether continuation of olanzapine beyond this period reduces relapse risk remains unclear but the potential benefit needs to be weighed against an increased risk of weight gain.Molecular Psychiatry advance online publication, 13 October 2015; doi:10.1038/mp.2015.158.