OBJECTIVE--To investigate if extracts of Hypericum perforatum (St John's wort) are more effective than placebo in the treatment of depression, are as effective as standard antidepressive treatment, and have fewer side effects than standard antidepressant drugs. DESIGN--Systematic review and meta-analysis of trials revealed by searches. TRIALS--23 randomised trials including a total of 1757 outpatients with mainly mild or moderately severe depressive disorders: 15 (14 testing single preparations and one a combination with other plant extracts) were placebo controlled, and eight (six testing single preparations and two combinations) compared hypericum with another drug treatment. MAIN OUTCOME MEASURES--A pooled estimate of the responder rate ratio (responder rate in treatment group/responder rate in control group), and numbers of patients reporting and dropping out for side effects. RESULTS--Hypericum extracts were significantly superior to placebo (ratio = 2.67; 95% confidence interval 1.78 to 4.01) and similarly effective as standard antidepressants (single preparations 1.10; 0.93 to 1.31, combinations 1.52; 0.78 to 2.94). There were two (0.8%) drop outs for side effects with hypericum and seven (3.0%) with standard antidepressant drugs. Side effects occurred in 50 (19.8%) patients on hypericum and 84 (52.8%) patients on standard antidepressants. CONCLUSION--There is evidence that extracts of hypericum are more effective than placebo for the treatment of mild to moderately severe depressive disorders. Further studies comparing extracts with standard antidepressants in well defined groups of patients and comparing different extracts and doses are needed.