To compare the safety and efficacy of paroxetine (n = 167) and placebo (n = 169), data from 4 centres using the same protocol were pooled. A double-blind parallel group design was used, with therapy lasting 6 weeks. Significant differences between paroxetine- and placebo-treated patients were found on the major efficacy outcome variables by week 2 and on all efficacy variables by week 4 of the study. Improvement on the sleep factor of the Hamilton Rating Scale for Depression was found after 7 d. Observer and patient global efficacy ratings were in agreement by week 4. No serious adverse events occurred, and paroxetine had no clinically significant effects on vital signs or laboratory safety data. Side effects were more common on paroxetine and were similar to other serotonin reuptake inhibitors. In general, these were well tolerated and did not lead to dropout. Symptoms of increased arousal were not seen during early therapy.