Previous investigations of the treatment of mania have resulted in uncertainty about the efficacy of lithium versus a neuroleptic. In addition there have been reports of toxicity with a haloperidol--lithium combination. In order to determine the comparative efficacy of lithium vs haloperidol vs a combination of haloperidol--lithium, we studied 21 severely ill manic patients who all met rigorous criteria for bipolar illness and who required in hospital treatment. Subjects were randomly assigned to 3 groups: (A) Lithium plus placebo (B) Placebo plus haloperidol and (C) Lithium plus haloperidol. The study was conducted in double blind fashion for 3 weeks with the dosages of the medications varied according to clinical response or untoward effects. Subjects on haloperidol and placebo or the haloperidol--lithium combination were significantly improved after 7 days in comparison to the lithium-treated group. Groups B and C did not differ from each other, either in degree of improvement or in side effects. Inspite of the relatively small sample size the results suggest (1) that haloperidol is superior to lithium for treating severely ill acute mania and (2) that while a haloperidol--lithium combination does not result in a significant increase in side effects, it is not superior to haloperidol alone.