Résumé : Background Few studies investigated socio-demographic and clinical predictors of non response and remission in treatment resistant depression (TRD) in the case of failure of more than two adequate antidepressant (AD) trial. The primary aim of this study was to investigate socio-demographic and clinical predictors of TRD defined as the lack of response to at least three adequate AD treatments, two of which prospectively evaluated. As secondary aims, we also investigated predictors of non response and remission to: (1) at least two adequate AD treatment (one of which prospectively assessed); (2) at least one adequate and retrospectively assessed AD treatment. Methods In the context of a European multicenter project, 407 major depressive disorder (MDD) patients who failed to respond to a previous AD treatment were recruited for a 2 stage trial, firstly receiving venlafaxine and then escitalopram. MINI, HRSD, MADRS, UKU, CGI-S and CGI-I were administered. Results Ninety eight subjects (27.61%) were considered as resistant to three AD treatments. Clinical predictors were: longer duration and higher severity of the current episode (p=0.004; ES=0.24; p=0.01; RR=1.41, respectively), outpatient status (p=0.04; RR=1.58), higher suicidal risk level (p=0.02; RR=1.49), higher rate of the first/second degree psychiatric antecedents (MDD and others) (p=0.04; RR=1.31, p=0.03; RR=1.32 respectively) and side effects during treatments (p=0.002; RR=2.82). Multivariate analyses underlined the association between TRD and the severity of the current episode (p=0.04). As for secondary outcomes, predicting factors were partially overlapping. Limitations The limited sample size and specific drugs used limit present findings. Conclusion Subjects with a high degree of resistance to AD treatments show specific features which may guide the clinicians to the choice of more appropriate therapies at baseline.