OBJECTIVE Few studies have compared simultaneously different antidepressants in long-term treatment of major depressive disorder (MDD). Long-term prevention of recurrences should be the main goal of MDD treatment. The purpose of this study was to compare antidepressants of different pharmacological classes in terms of retention in treatment (no discontinuation for recurrences, hospitalizations, side effects). METHODS One hundred and fifty outpatients with an MDD diagnosis, treated with antidepressants in mono-therapy, were included. Follow-up period was set at 24 months, and information have been obtained from charts, interviews with patients and their relatives, and from the Lombardy regional register. A survival analysis (Kaplan-Meier) was performed, considering recurrences, hospitalizations, or discontinuation due to side effects as 'death' events. RESULTS In our sample, 48.7% of the patients presented a recurrence within the first 2 years of treatment. Bupropion appears less effective in long-term treatment of MDD than the other compared antidepressants, with exception of fluoxetine (p = 0.09), amitriptyline (p = 0.13), fluvoxamine (p = 0.83), venlafaxine (p = 0.5), and trazodone (p = 0.58). Fluvoxamine appears to be less effective than citalopram (p = 0.036), paroxetine (p = 0.037), clomipramine (p = 0.05), sertraline (p = 0.011), and duloxetine (p = 0.024). CONCLUSIONS Bupropion and fluvoxamine appear less effective in long-term treatment of MDD. These results should be confirmed by randomized placebo-controlled prospective studies with larger samples.