OBJECTIVES: Compare adults treated with SSRIs and SNRIs for major depression. METHODS: Identified all head-to-head trials comparing SSRIs (citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, sertraline) with SNRIs (venlafaxine-XR, duloxetine) in therapeutic doses. Outcome: remission (HAMD-7 ?3 or HAMD-17 ?7 or MADRS ?12) at 8 weeks. Two reviewers searched Medline, Embase and Cochrane databases to identify articles, extract data, adjudicated by a third judge. Rates were combined using random-effects meta-analytic model. Performed Intent-to-Treat (ITT) and Per-Protocol (PP) analyses. ?2 assessed heterogeneity of effects. RESULTS: 25 studies were identified, 19 were rejected; 6 studies provided 7 head-to-head trials of 1345 patients (68.0% females per drug). Five RCTs (N = 1008) and 2 naturalistic trials (N = 337). All displayed non-heterogeneity (P > 0.05). ITT Remission rates in RCTs: 49.5% (SE = 6.2%, n = 398) for SNRIs; 39.3% (SE = 10.0%, n = 369) for SSRIs; meta-analytic difference 9.2% (CI95%:3.0%-15.4%). PP rates - 67.8% (SE = 7.5%, n = 297) for SNRIs; 56.5% (SE = 10.9%, n = 269) for SSRIs; meta-analytic difference 9.8% (CI95%:0.2%-19.5%). Naturalistic studies produced ITT rates of 37.6% (SE = 3.6%, n = 178) for SNRIs; 40.2% (SE = 6.7%, n = 159) for SSRIs; a non-significant (P = 0.69) difference of 2.1% favoring SSRIs. PP rates 46.2% (SE = 4.1%, n = 145) for SNRIs; 49.6% (SE = 9.0%, n = 129) for SSRIs; a difference of 2.9% (P = 0.68) favoring SSRIs. ITT remission rates were 46.2% (SE = 5.1%, n = 737) for SNRIs; 39.5% (SE = 7.2%, n = 608) for SSRIs; meta-analytic difference of 6.5% (CI95%:0.2% - 12.8%). PP rates were 61.8% (SE = 6.9%, n = 571) and 54.5% (SE = 8.1%, n = 450); meta-difference was 6.4% (P = 0.13). CONCLUSION: SNRIs seem more efficacious. Naturalistic studies produced non-significant results differing from RCT results.