OBJECTIVE: This study evaluated the efficacy and tolerability of escitalopram and duloxetine in the treatment of major depressive disorder (MDD). RESEARCH DESIGN AND METHODS: Patients were randomised to 24 weeks of double-blind treatment with fixed doses of escitalopram (20 mg) (n = 143) or duloxetine (60 mg) (n = 151). The primary analysis of efficacy was an analysis of covariance (ANCOVA) of change from baseline to endpoint (week 24) in MADRS total score (last observation carried forward). MAIN OUTCOME MEASURES; RESULTS: At week 8, the mean change from baseline in total MADRS score was -19.5 for patients treated with escitalopram (n = 141) and -17.4 for patients treated with duloxetine (n = 146), a difference of 2.1 points (p < 0.05). At week 8, the proportion of responders (> or = 50% decrease in MADRS) was 69% (escitalopram) and 58% (duloxetine) (p < 0.05) and remission (MADRS < or = 12) rates were 56% (escitalopram) and 48% (duloxetine) (NS). For the primary endpoint, the mean change from baseline in total MADRS score at week 24 was -23.4 for patients treated with escitalopram and -21.7 for patients treated with duloxetine, a difference of 1.7 points (p = 0.055, one-sided). The difference in mean change from baseline in MADRS total score favoured escitalopram at weeks 1, 2, 4, 8, 12 and 16 (p < 0.05). The overall withdrawal rates were 22% (escitalopram) and 25% (duloxetine) (NS). The withdrawal rate due to adverse events was lower for escitalopram (9%) compared to duloxetine (17%) (p < 0.05) and significantly more patients treated with duloxetine reported insomnia (12.6% vs. 4.9%) and constipation (8.6% vs. 2.8%). CONCLUSION: Escitalopram was superior to duloxetine in acute treatment and at least as efficacious and better tolerated in long-term treatment of MDD.