Forty-five patients with acute schizophrenic illnesses (defined by PSE criteria) were assessed in clinical, cognitive and social terms before being entered in a four week study of the isomers of flupenthixol and placebo. At the end of one year they were re-assessed in the same terms. The clinical and psychological features of the acute illness and the drug treatment given did not predict outcome. Poor outcome in social terms was significantly related to severe social isolation in the initial assessment and to the presence of nuclear symptoms and negative schizophrenic features at follow-up.