The hypothesis that lithium is the most appropriate and efficacious pharmacological treatment in the therapy of manic-depressive illness was examined in a meta-analysis. The efficacy of lithium was compared with that of the antiepileptics, carbamazepine and valproic acid. Data sources include the Medline database and relevant references from articles. Only randomised controlled clinical trials with a double-blind assessment of outcomes were included. Lithium levels were required to be within the therapeutic range of 0.4-1.5 mmol/l. The effect sizes were measured by the odds ratio using the Mantel-Haenszel method. No significant difference was observed between the treatment effect of the 3 drugs compared. While the efficacy of the 3 treatments was identical, there is a general tendency in favour of the antiepileptics in the treatment of manic depression with regard to adverse events and treatment tolerance. The impact and importance of this study seriously question the relevance of today's practice of prescribing lithium in the acute management of manic-depressive illness.