Thirty-six acute schizophrenics were randomly assigned to dextro (d)-propranolol or placebo in a double blind trial lasting four weeks. All patients had a fixed dose of haloperidol during the first week, which resulted in an initial improvement in both groups. Thereafter, a deterioration towards base-line was seen. Six patients on placebo, but none on propranolol were treatment failures at the end of three weeks (P < 0.001). Comparison of change in scores from week 2 to 4 showed significantly greater deterioration in the placebo group; d-propranolol thus had a better effect than placebo in sustaining the initial improvement with haloperidol. The overall magnitude of clinical change from pre-treatment scores is small, the majority of the patients showing little or no overall improvement. It is concluded that d-propranolol has a detectable therapeutic effect, which by inference must have a novel pharmacological basis, but this is not as potent as standard neuroleptics.