OBJECTIVE The effects of olanzapine and divalproex for the treatment of mania were compared in a large randomized clinical trial. METHODS A 3-week, randomized, double-blind trial compared flexibly dosed olanzapine (5-20 mg/day) to divalproex (500-2500 mg/day in divided doses) for the treatment of patients hospitalized for acute bipolar manic or mixed episodes. The Young Mania Rating Scale and the Hamilton Depression Rating Scale were used to quantify manic and depressive symptoms, respectively. Safety was assessed with several measures. RESULTS The protocol defined baseline-to-endpoint improvement in the mean total score on the Young Mania Rating Scale as the primary outcome variable. The mean Young Mania Rating Scale score decreased by 13.4 for patients treated with olanzapine (N=125) and 10.4 for those treated with divalproex (N=123). A priori categorizations defined response and remission rates: 54.4% of olanzapine-treated patients responded (> or = 50% reduction in Young Mania Rating Scale score), compared to 42.3% of divalproex-treated patients; 47.2% of olanzapine-treated patients had remission of mania symptoms (endpoint Young Mania Rating Scale < or = 12), compared to 34.1% of divalproex-treated patients. The decrease in Hamilton depression scale score was similar in the two treatment groups. Completion rates for the 3-week study were similar in both groups. The most common treatment-emergent adverse events (incidence >10%) occurring more frequently during treatment with olanzapine were dry mouth, increased appetite, and somnolence. For divalproex, nausea was more frequently observed. The average weight gain with olanzapine treatment was 2.5 kg, compared to 0.9 kg with divalproex treatment. CONCLUSIONS The olanzapine treatment group had significantly greater mean improvement of mania ratings and a significantly greater proportion of patients achieving protocol-defined remission, compared with the divalproex treatment group. Significantly more weight gain and cases of dry mouth, increased appetite, and somnolence were reported with olanzapine, while more cases of nausea were reported with divalproex.