Antidepressant effects and unintended effects of paroxetine (30 mg/day) and clomipramine (150 mg/day) were compared in a double-blind, randomized, inpatient, fixed-dose, plasma-level-controlled study. Patients with a DSM-III diagnosis of major depressive disorder participated. After 1 week of singfe-blind placebo treatment 120 patients fulfilled the criterion of a Hamilton (17-item) score of ≥ 18 and were started on active treatment for 6 weeks. Drop-outs on paroxetine (n = 12) were largely due to lack of effect, and on clomipramine (n = 19) due to lack of effect (n = 7), adverse reactions or severe side effects (n = 10) and development of mania (n = 2). According to the protocol, non-responders (Hamilton total ≥ 16) after 4 weeks active treatment were terminated, and this occurred to 23 patients on paroxetine and four patients on clomipramine. Categorical response measures and group averages of rating scores showed a significantly better therapeutic effect of clomipramine from the second week of treatment on. These results are very similar to our earlier results with another selective serotonin reuptake inhibitor (citalopram), but generally at variance with the literature on this class of antidepressants, which, however, mostly deals with outpatient studies.