This meta-analysis examined the effectiveness of treatments of bipolar depression. Trials were identified using the MEDLINE, EMBASE, http://www.clinicaltrials.gov, and Cochrane databases (1993 to July 2008). The outcome measures included mean change in Montgomery-Asberg Depression Rating Scale (MADRS) or Hamilton Depression Rating Scale (HAM-D) total scores, and rates of response and remission. Overall, 19 publications were included. Medications included quetiapine, lamotrigine, paroxetine, lithium, olanzapine, aripiprazole, phenelzine, and divalproex. The most trials were identified for quetiapine (5) and lamotrigine (6). Not all medications were associated with symptomatic improvement (significant reduction in MADRS/HAM-D total scores vs placebo), with lamotrigine, paroxetine, aripiprazole, and lithium not being different from placebo. Highest reductions in MADRS scores versus placebo were reported for the olanzapine-fluoxetine combination (1 trial: -6.6; 95% confidence interval [CI], -9.59 to -3.61; P = 0.000) and quetiapine monotherapy (5 trials: for 300 mg/d, -4.8; 95% CI, -6.18 to -3.49; P = 0.000; for 600 mg/d, -4.8; 95% CI, -6.22 to -3.28; P = 0.000), with quetiapine monotherapy also showing the highest reduction in HAM-D scores (4 trials: -4.0; 95% CI, -5.0 to -2.9; P = 0.000). All medications except paroxetine, lithium, aripiprazole, and phenelzine significantly improved the ratio of probabilities of response (overall rate, 1.31; 95% CI, 1.22-1.40) and remission (1.32; 95% CI, 1.20-1.45) versus placebo. Variability in efficacy exists between treatments of bipolar depression. Quetiapine and the olanzapine-fluoxetine combination showed the greatest symptomatic improvement. Efficacy considerations will need to be balanced against safety and tolerability of the individual agents.