Two recent large-scale studies of the effectiveness and safety of divalproex versus lithium and placebo in the manic phase of bipolar disorder have provided an opportunity to investigate the characteristics of patients enrolled in clinical drug trials of mood stabilizers. Both studies are randomized, double-blind, parallel-group, multicenter trials; the first addresses effectiveness in acute mania, and the second addresses effectiveness in prevention of manic episodes. Data on demographic, illness severity, and treatment response characteristics were compared between Study 1 and six other clinical drug trials in acute mania from the past decade that had data in comparable domains. Although patients experienced onset of bipolar disorder in their early twenties, the average age of enrollment in the studies was about 40. Approximately equal numbers of men and women were enrolled. The patients were severely ill at the time of enrollment, with mean scores on the Global Assessment Scale ranging from 20 to 37. Overall, characteristics were remarkably similar across studies and representative of those reported in epidemiological studies. From Study 2, which is currently in progress, characteristics of patients recruited into the trial were determined. At least 10 subjects were initially assessed for each subject actually enrolled in the trial. The two factors most associated with failure to move from the open phase of the study to the randomized phase were administrative problems (principally failure of subjects to adhere to the medication regimen and protocol requirements) and inability, even with an open treatment protocol, to achieve adequate recovery from the index manic episode within 3 months.