In a randomized clinical trial with an observation period of 2.5 years, the differential efficacy of lithium versus carbamazepine was compared in 171 bipolar patients (DSM-IV). In order to investigate the efficacy of the two drugs in clearly defined subsamples, a series of subgroup analyses was carried out. First, patients with a bipolar I disorder (n = 114) were analyzed separately. In these patients, lithium was superior to carbamazepine. In contrast, carbamazepine was at least equally as efficacious as lithium in the subsample of patients with bipolar II disorder or bipolar disorder not otherwise specified (n = 57). In a second analysis on differential efficacy, the whole sample was subdivided into a classical subgroup (bipolar I patients without mood-incongruent delusions and without comorbidity; n = 67) and a nonclassical subgroup including all other patients (n = 104). Classical bipolar patients had a significantly lower hospitalization rate under lithium than under carbamazepine prophylaxis (26 vs. 62%, p = 0.012). For the nonclassical group, a tendency in favor of carbamazepine was found. In a third step, we analyzed the impact of episode sequence on differential efficacy. In a global view, the episode sequence prior to the index episode was not correlated to differential efficacy. Our results might, however, indicate that patients with an episode sequence of mania-depression-free interval responded better to lithium. Besides differential efficacy, suicidal behavior and patients' satisfaction with treatment were investigated. Regarding suicidal behavior, a trend in favor of lithium was found. The data on patients' satisfaction were significantly in favor of carbamazepine. In conclusion, lithium appears to be superior to carbamazepine in classical bipolar cases and might have additional impact on proneness to suicide. The distinctly larger group of patients with nonclassical features might profit more from carbamazepine which seems to be well accepted by the patients. Hence, treatment alternatives to lithium a re desirable for the majority of bipolar patients. Copyright (C) 2000 S. Karger AG, Basel.