Fluoxetine was developed as an antidepressant drug. It is more effective than placebo, but a dose-effect relation has not been established. Fluoxetine is almost as effective as tricyclic antidepressant drugs, but the available studies do not allow accurate comparisons. Fluoxetine may be less effective than tricyclic antidepressant drugs for the treatment of inpatients with severe melancholic depression, and it should not be the first choice of a drug for them. Fluoxetine may be most appropriate for patients with moderate depression who can be treated as outpatients. If there is little improvement after treatment for four to six weeks, an alternative treatment should be offered. Fluoxetine does not have the anticholinergic, hypotensive, and sedative effects of tricyclic antidepressant drugs and has no particular cardiovascular effects; overdoses do not cause serious toxic effects. Nausea, anorexia, insomnia, and nervousness--the most common side effects--may be controlled with a careful adjustment to the dose. Clinically important drug interactions may occur with monoamine oxidase inhibitors, tricyclic antidepressant drugs, and other drugs. The published data on the antidepressant effect of fluoxetine do not fully explain its popularity. One may speculate that fluoxetine has psychobiologic effects not strictly related to the biology of depression and that it acts primarily as a mood- or affect-modulating agent.