BACKGROUND Few reliable correlates of treatment response in depression have emerged despite nearly 40 years of research. We examined the correlates of recovery in a "mega-analysis," or meta-analysis of original data, of 595 patients with major depressive disorder enrolled in 6 standardized treatment protocols. METHODS All patients (mean age, 44 years; 31% male and 69% female) met criteria for nonbipolar, nonpsychotic primary major depressive disorder and were treated for 16 weeks with either cognitive behavior therapy or interpersonal psychotherapy alone (psychotherapy alone; n = 243) or interpersonal psychotherapy plus antidepressant pharmacotherapy (combined therapy; n = 352). The impact of treatment type, severity, study, and other covariates on recovery rates or time to recovery were examined by means of chi 2, log-rank tests, the Cox proportional hazards model, and sensitivity analyses. RESULTS Whereas combined therapy was not significantly more effective than psychotherapy alone in milder depressions, a highly significant advantage was observed in more severe recurrent depressions. Poorer outcomes were also observed in women and older patients, although these effects were dependent on inclusion of particular studies. CONCLUSIONS Mega-analysis is a powerful method for comparing the efficacy of treatments and examining correlates of response. Using this method, we found new evidence in support of the widespread clinical impression that combined therapy is superior to psychotherapy alone for treatment of more severe, recurrent depressions.