OBJECTIVE: The authors sought to determine whether the manic/mixed episode distinction in patients with bipolar disorder runs true over time. METHOD: Over an 11-year period, the observed distribution of manic and mixed episodes (N=1,224) for patients with three or more entries in the management information system of a community mental health center (N=241) was compared to the expected distribution determined by averaging 1,000 randomly generated simulations. RESULTS: Episodes were consistent (all manic or all mixed) in significantly more patients than would be expected by chance. CONCLUSIONS: These data suggest a pattern of diagnostic stability over time for manic and mixed episodes in patients with bipolar disorder. Careful prospective studies of this issue are needed.