Objective: To determine the short-term antidepressant efficacy and tolerability of duloxetine and venlafaxine vs. each other, placebo, selective serotonin reuptake inhibitors (SSRIs), and tri- and tetracyclic antidepressants (TCAs) in adults with major depression. Method: Meta-analysis of randomised controlled trials identified through bibliographical databases and other sources, including unpublished manufacturer reports. Results: Fifty-four studies including venlafaxine arms (n=12816), 14 including duloxetine arms (n=4528), and two direct comparisons (n=836) were analysed. Twenty-three studies were previously unpublished. In the meta-analysis, both duloxetine and venlafaxine showed superior efficacy (higher remission and response rates) and inferior tolerability (higher discontinuation rates due to adverse events) to placebo. Venlafaxine had superior efficacy in response rates but inferior tolerability to SSRIs (OR=1.20, 95% CI 1.07-1.35 and 1.38, 95% CI 1.15-1.66, respectively), and no differences in efficacy and tolerability to TCAs. Duloxetine did not show any advantages over other antidepressants and was less well tolerated than SSRIs and venlafaxine (OR=1.53, 95% CI 1.10-2.13 and OR 1.79, 95% CI 1.16-2.78, respectively). Conclusion: Rather than being a first-line option, venlafaxine appears to be a valid alternative in patients who do not tolerate or respond to SSRIs or TCAs. Duloxetine does not seem to be indicated as a first-line treatment.