In a study of 18 patients with manic symptomatology and 31 patients with melancholic symptomatology the Bech-Rafaelsen Mania Scale (BRMS) and the Hamilton Depression Scale (HDS) have been compared. The results showed that the inter-observer reliability of the BRMS was adequate compared with the HDS. Both scales are constructed for assessing the severity of manic or melancholic states, and no difference was found in the total BRMS or HDS score between the various diagnostic groups, when the patients were classified by an index of the course and symptomatology of their disorder, using the Multi-axial Classification System for Affective Disorders (MULTI-CLAD). The homogeneity of the BRMS seemed more adequate than that of the HDS, when each item was correlated to the corresponding total score. Although the homogeneity of the BRMS needs to be evaluated by other statistical models than correlation analysis, our results seem to indicate that the improvement in assessing manic-melancholic states quantitatively is a matter of redefining items or incorporating new items in the melancholic rather than the manic part of these rating scales.