Reports on the effects of benzodiazepines in schizophrenia have appeared since the early 1960s. Conclusions drawn from these studies, most of which have been uncontrolled, have ranged from worse than placebo to better than neuroleptics. A critical appraisal of the literature seems to warrant the following main conclusions. Benzodiazepines alone, in conventional doses, have no convincing antipsychotic effect in schizophrenia, although they may reduce anxiety, tension and insomnia. However, very high doses of diazepam, and possibly other benzodiazepines, may have a symptomatic antipsychotic effect, especially in paranoid-hallucinatory schizophrenics, also when given alone. Benzodiazepines, in conventional doses, can enhance the antipsychotic effect of neuroleptics in schizophrenics who have not responded satisfactorily to neuroleptics alone. This effect may be most conspicuous against hallucinations, but improvement may also be obtained from delusions, thought disturbances, some negative symptoms, anxiety and tension. Some benzodiazepines may be more effective than others in schizophrenia, but this has been insufficiently elucidated.