Résumé : OBJECTIVES: Very few studies have investigated clinical features associated with treatment-resistant depression (TRD) defined as failure of at least 2 consecutive antidepressant trials. The primary objective of this multicenter study was to identify specific clinical and demographic factors associated with TRD in a large sample of patients with major depressive episodes that failed to reach response or remission after at least 2 consecutive adequate antidepressant treatments. METHOD: A total of 702 patients with DSM-IV major depressive disorder, recruited from January 2000 to February 2004, were included in the analysis. Among them, 346 patients were considered as nonresistant. The remaining 356 patients were considered as resistant, with a 17-item Hamilton Rating Scale for Depression score remaining greater than or equal to 17 after 2 consecutive adequate antidepressant trials. Cox regression models were used to examine the association between individual clinical variables and TRD. RESULTS: Among the clinical features investigated, 11 variables were found to be associated with TRD. We found anxiety comorbidity (p < .001, odds ratio [OR] = 2.6), comorbid panic disorder (p < .001, OR = 3.2) and social phobia (p = .008, OR = 2.1), personality disorder (p = .049, OR = 1.7), suicidal risk (p = .001, OR = 2.2), severity (p = .001, OR = 1.7), melancholia (p = .018, OR = 1.5), a number of hospitalizations > 1 (p = .003, OR = 1.6), recurrent episodes (p = .009, OR = 1.5), early age at onset (p = .009, OR = 2.0), and nonresponse to the first antidepressant received lifetime (p = .019, OR = 1.6) to be the factors associated with TRD. CONCLUSIONS: Our findings provide a set of 11 relevant clinical variables associated with treatment resistance in major depressive disorder that can be explored at the clinical level. The statistical model used in this analysis allowed for a hierarchy of these variables (based on the OR) showing that comorbid anxiety disorder is the most powerful clinical factor associated with TRD.