There are currently seven rating scales available to assess manic symptomatology. All, however, have some limitations that could restrict their clinical and research utility. To resolve these deficiencies the Clinician-Administered Rating Scale for Mania (CARS-M) was developed and normed on 96 patients with mixed diagnoses during baseline and following treatment. Interrater reliability was established across multiple raters viewing 14 videotaped interviews and comparing agreement among individual items and total scores. Test-retest reliability was assessed on 36 patients twice during baseline. The mean intraclass correlation coefficient among five raters across items for each of the 14 patients was 0.81, and for total scores 0.93. Principal components analysis of items revealed two factors: mania, and psychosis. Test-retest reliability was significant for both factors (range = 0.78 to 0.95). Internal validity, comparing each item with its respective total factor score, revealed significant correlations for all items. Correlation of CARS-M total scores with mania rating scale (MRS) total scores was 0.94. Results indicate the CARS-M is both a reliable and valid measure of the severity of manic symptomatology, which incorporates a number of methodological improvements leading to greater precision and clinical utility.