Recurrent unipolar depression is a common, but undertreated disorder. Many patients require long-term maintenance therapy, and full doses of antidepressant agents may be preferred for the prevention of relapse. We report results of a 1-year, multicenter, open-label study of paroxetine (10 to 50 mg/day) in 433 patients with major depressive disorder, with additional data from 110 patients who entered a long-term extension of the study. The primary measures of efficacy were the Hamilton Rating Scale for Depression (HAM-D) total and Clinical Global Impression (CGI) severity of illness scores. During the first 6 weeks of therapy, the mean HAM-D total declined approximately 50% (from 27.9 to 13.5), with continued improvement, at an attenuated rate, throughout the first year. At the end of 1 year, the mean HAM-D total was 6.9. Similarly, the CGI severity of illness score declined from 4.6 at baseline to 2.8 at week 6 and to 1.7 at the end of 1 year. Remission was maintained in the population that entered the long-term extension, with mean HAM-D total and CGI severity of illness scores of 6.4 and 1.8, respectively, after 2.5 years, and 4.2 and 1.3 after 4 years. The most common adverse events reported during long-term treatment with paroxetine were somnolence, nausea, headache, and sweating. Pharmacokinetic analysis showed no clear correlation between the concentrations of paroxetine in plasma and either clinical efficacy or tolerability. There was no increased drug accumulation during long-term treatment. Side effects tended to occur early during therapy; and no new side effects emerged during the long-term extension. These results suggest that paroxetine is effective and well tolerated in the long-term treatment of depression.