The purpose of this study was to compare the safety and efficacy of a relatively new antidepressant drug, alprazolam (a triazolobenzodiazepine) with imipramine in the treatment of 60 depressed symptomatic volunteers. Eligible patients were randomly assigned after a 1-week washout to one of the medications and followed for 6 treatment weeks. Contrary to the earlier report of Feighner et al. (1983), who found alprazolam superior to imipramine and placebo, but consistent with Rush et al. (1985) we find imipramine superior in efficacy to alprazolam on a variety of symptoms. Both the present study and Rush's study employed patients with signs indicative of response to tricyclics. Feighner's patients may have been the type who tend to be less responsive to tricyclics but may be more responsive to alprazolam. Some of our data also show that alprazolam may have a more advantageous effect in the early weeks of treatment but is overtaken in subsequent treatment weeks by imipramine.