Depression is associated with considerable morbidity and mortality. As depressive disorders carry a high risk of relapse, treatment strategies include the use of a 6-month continuation period after resolution of the acute episode. Tolerability is of major importance when determining compliance and outcome during long-term therapy. Due to the superior tolerability profile of the selective serotonin reuptake inhibitors (SSRIs) over the older tricyclic antidepressants (TCAs), the former may be more suitable for extended therapy. Comparative studies have not shown differences between the SSRIs in terms of efficacy, but side-effect profiles may vary. A multicenter, double-blind, comparative study of sertraline and fluoxetine was carried out in outpatients fulfilling DSM-III-R criteria for major depressive disorder. Patients were randomised to receive sertraline (50—150 mg, n = 118) or fluoxetine (20—60 mg, n = 120) for 24 weeks. Assessments for depression (HAM-D, HAD, CGI-I, CGI-S), anxiety (Covi), sleep (Leeds Sleep Evaluation scale) and quality of life (SIP) were made at study entry and at weeks 2, 4, 8, 12, 18 and 24. All adverse events were recorded to allow evaluation of tolerability. In total, 88 patients in the sertraline group completed the study compared with 79 in the fluoxetine group. Side effects were responsible for the premature treatment withdrawal of seven (6%) sertraline patients and 12 (10%) fluoxetine patients. Two-hundred and thirty-four patients were included in an ITT analysis up to last visit (116 sertraline, 118 fluoxetine). At study endpoint, both treatments produced a significant improvement over baseline on all efficacy variables (P < 0.001). Although the magnitude of global changes in depression, anxiety, and quality of life was larger with sertraline than fluoxetine, none of the between-group differences reached statistical significance. However, significant differences in favour of sertraline were observed for individual HAM-D items including item 4 (insomnia onset) (P = 0.04), item 9 (agitation) (P = 0.02), and item 13 (general somatic symptoms) (P = 0.008). In addition, sertraline was associated with significantly superior performance on the Leeds Sleep Evaluation scale and on SIP items relating to sleep and rest, emotional behaviour and ambulation. Both sertraline and fluoxetine were well tolerated with no significant differences between treatments.