The herbal remedy St John's wort is widely used as an antidepressant but its efficacy has not been systematically investigated. Meta-analyses and systematic reviews of published trials strongly suggest St John's wort is more effective than placebo although comparative efficacy to standard antidepressants is less clearly established. We updated and expanded previous meta-analyses of St John's wort, scrutinised the validity of published reports and examined possible mechanisms of action. Twenty-two randomised controlled trials were identified. Meta-analysis showed St John's wort to be significantly more effective than placebo (relative risk (RR) 1.98 (95% CI 1.49-2.62)) but not significantly different in efficacy from active antidepressants (RR 1.0 (0.90-1.11)). A sub-analysis of six placebo-controlled trials and four active comparator trials satisfying stricter methodological criteria also suggested that St John's wort was more effective than placebo (RR 1.77 (1.16-2.70)) and of similar effectiveness to standard antidepressants (RR 1.04 (0.94-1.15)). There was no evidence of publication bias. Adverse effects occurred more frequently with standard antidepressants than with St John's wort. The mechanism of action of St John's wort remains unknown. Future research should include large scale, appropriately powered comparisons of St John's wort and standard antidepressants.