Seventy patients with unipolar major depressive disorder were treated with fluoxetine or placebo in a 6-week double-blind trial and were evaluated by changes in scores on the Hamilton Rating Scale for Depression (HAM-D) and the global improvement measure of the Clinical Global Impressions (CGI) scale. High correlations were found between the changes in HAM-D scores from baseline to endpoint and the final CGI improvement ratings. In patients with moderate depression (baseline HAM-D score of 20 or more), the differences in endpoint analysis between active treatment and placebo groups were significant. A persistent pattern of improvement was noted in 27% of those receiving fluoxetine but in none of those receiving placebo. Physician and patient evaluations as determined by the improvement measure of the CGI were closely correlated.