OBJECTIVES: Clinical registration studies in Major Depressive Disorder (MDD) are mostly controlled versus placebo. Regulators, clinicians, patients and payers therefore lack robust information to compare a new drug to the alternatives at time of launch. Objective was to evaluate if relative antidepressant efficacy varies over time. We evaluated the evolution of escitalopram and citalopram. METHODS: A database of randomized-controlled trials (RCTs) of adults suffering from MDD was built in order to provide a comprehensive assessment of antidepressant efficacy. New-generation antidepressants launched from 1989 to 2002 and placebo were included in this analysis. A 5-year period was chosen for the efficacy evaluation (from 2002 to 2007). Analyses were performed on the Montgomery-Asberg Depression Scale (MADRS) adjusted mean change from baseline at 2 months (6-12 weeks) using a Bayesian Mixed Treatment Comparison method with random effect and Normal likelihood. Escitalopram efficacy evolution was presented as mean difference to placebo, ranking probabilities and mean rank. RESULTS: MADRS results were reported at 2 months in 122 RCTs; 83 were selected for this analysis (excluding treatments launched after 2002). Differences in MADRS total score versus placebo increased from -3.39 [-5.06;-1.70] in 2002 to -3.76 [-4.63;-2.90] in 2007 for escitalopram. Ranking probabilities curves for escitalopram and citalopram were mostly overlapping in 2002, while a much clearer separation in favour of escitalopram appeared in 2007. Mean ranks were respectively 6.8 and 6.2 in 2002; 4.4 and 7.3 in 2007 for escitalopram and citalopram. CONCLUSIONS: Escitalopram relative efficacy increased from 2002 to 2007. This was mainly explained by new positive superiority escitalopram studies. Time of launch did not appear always to be the most appropriate to assess antidepressant efficacy mostly based on RCTs versus placebo. Other outcomes and studies selection may have an impact on results.