Background: Recent investigations have suggested that the antimanic agents divalproex sodium and carbamazepine may each hasten hospital discharge and be especially beneficial in treating mixed-state mania. This study retrospectively compared the time to remission for pure versus mixed manic bipolar inpatients who were taking lithium, divalproex, or carbamazepine, or their combination, under naturalistic conditions. Method: Records were reviewed for 120 bipolar inpatients from 1991 to 1995. Research DSM-III-R diagnoses of pure or mixed mania were assigned along standardized guidelines. Data were obtained on daily symptoms, medication doses, and blood levels. Weekly improvement was evaluated by Kaplan-Meier survival analysis of Clinical Global Impressions scale scores. Variables associated with 'remission' versus 'nonremission' were examined by logistic regression. Results: Mixed mania (N = 70) was more common than pure mania (N = 50). No significant differences were observed in the time to remission for mixed or pure manic bipolar patients who took lithium compared with those who took divalproex or carbamazepine. In patients who remained symptomatic with lithium as a single-agent mood stabilizer despite therapeutic serum lithium levels, the addition of a second mood stabilizer led to rapid symptom improvement. Among all medication subgroups, the speed with which patients achieved therapeutic blood levels of any of these agents significantly affected the time to remission. Conclusion: Mixed manic bipolar patients taking lithium, divalproex, or carbamazepine under naturalistic conditions remit at comparable rates. Those failing to respond to single-agent mood stabilizers often receive combinations of mood stabilizers. However, delays in optimizing a medication regimen may attenuate short-term outcome, regardless of the mood stabilizer selected. Rapid achievement of therapeutic blood levels of any antimanic agent appears to be strongly related to swift symptom remission.