As a measure of severity, the Hamilton scale has frequently been tested against clinical ratings of severity. The results are generally good, although variable. The total score of the 17-item version was significantly related to globally assessed severity of depression on the Raskin three-item rating scale for depression (r= 0.65); correlations for the Montgomery-Åsberg Depression Rating Scale (MADRS) (0.71) and BRMS (0.70) were slightly higher (26, Tables 1–3). Four items in the Hamilton scale were not significantly related to the global rating. Other figures include a correlation of 0.81 between the HRDS and the Raskin scale, and -0.86 between the HRDS and overall severity rated by the Global Assessment Scale (GAS); correlations between individual items and the global ratings ranged from 0.10 to 0.86 (11, Table 3). Hedlund and Vieweg’s review cited correlations of 0.84, 0.89, and 0.90 with clinical ratings of severity; they also listed a lower correlation of 0.67 for a study that included depressed and nondepressed patients (12, p152). Other low correlations with a clinical severity rating include 0.68 (16, Table 3). The HRSD may be limited in its ability to distinguish moderate and severe depression (16; 31), and the results given here indicate that not all the items correlate with global assessments.