This study examines the predicting factors for absenteeism in depressed patients. Using a 'cross-sectional' design, we observed 345 patients diagnosed with major depressive disorders as assessed by the Diagnostic and Statistical Manual for Mental Disorders, third edition revised (DSM-III-R) criteria and Hamilton Depression Rating Scale (HAM-D) [12] score higher than 12. The treatment group (n = 268) were treated with antidepressants (n = 98 with fluoxetine and n = 170 with tricyclics [amitriptyline, clomipramine]) for at least one week and the non treated group (n = 67) had not received antidepressants for at least one month. Sociodemographic, clinical and therapeutic data was collected. The primary endpoint was absenteeism from work. Logistic regression analysis of these data was used to identify potential predictive variables. The rate of absenteeism from work was greater in non treated (70.2%) compared to treated patients (39.8% for fluoxetine group and 57.7% for tricyclics group). The risk of absenteeism for patients treated with tricyclics was 2.45 times greater than for patients treated with fluoxetine (odds-ratio = 2.45, CI 95% = 1.1-4.7). For all patients, the strongest predictors of absenteeism from work were symptom severity (odds-ratio = 44.4, CI 95% = 7.9-250) followed by past history of depression (odds-ratio = 6.85, CI 95% = 2.6-18.4) and past history of absenteeism (odds-ratio = 6.51, CI 95% = 2.0-204). In conclusion, the risk of absenteeism from work increases with depression severity and is higher with tricyclics compared to fluoxetine.