This double blind study assessed the antidepressant properties of diazepam, a minor tranquilizer, phenobarbital sodium, a barbiturate, and amitriptyline, a tricyclic antidepressant, in 138 nonpsychotic depressed and mixed anxious depressed outpatients. Its major purpose was to examine whether the frequently claimed antidepressant effects of minor and major tranquilizers are specific for these agents or may simply be explained by their sedative effects. Patients in this study, treated in general practice, private psychiatric practice, or a psychiatric clinic, responded as well to the primarily sedating phenobarbital sodium (120 mg/day) as to either diazepam (20 mg/day) or amitriptyline (100 mg/day). This suggests that antidepressant effects may indeed be largely explained by sedative effects.Such nondrug factors as treatment setting, severity of initial anxiety and depression, and diagnostic category did not differentially affect response to the three agents. There was evidence, however, that phenobarbital produced more improvement in patients with lower educational levels and greater family stress, confirming earlier results showing such patients to respond particularly well to the primarily sedative effects of this agent. Also, diazepam produced more improvement in patients who reported side effects than in patients who did not, suggesting that the present diazepam dosage may not have been high enough to exert maximal antidepressant effects.Amitriptyline was associated with more attrition, dosage deviation, autonomic nervous system side effects, and severe side effects than the other two agents. Diazepam produced the least attrition and greatest number of sedative side effects. Drug differences in attrition, dosage deviation, and side effects did not, however, reach statistical significance.The lack of main drug effects in the present study cannot be attributed to a low improvement rate. Indeed, evidence from several sources was presented to indicate that all three drugs produced a relatively high degree of clinical improvement over the 4- to 6-week treatment period.