Thirty-four manic patients were randomly assigned to treatment with carbamazepine or lithium. Clinical response was rated over 4 weeks. Twenty-eight patients, 14 in each group, completed the full protocol. Serum levels for both drugs were within the accepted therapeutic range. The overall response to treatment was not significantly different between the two groups. Comparison of individual Clinical Global Impressions, Brief Psychiatric Rating Scale, and Beigel-Murphy Manic State Rating Scale change scores showed a more consistent level of improvement across patients in the lithium-treated group compared to a minority of good responders in the carbamazepine-treated group. The findings suggest that carbamazepine has antimanic potential in specific bipolar patients whose clinical characteristics remain to be clearly defined.