In a seven-hospital collaborative study, 210 chronic schizophrenics were assigned to a placebo for a 24-week period. During that time, 40 per cent. of the patients relapsed and had to be returned to active medication. Probability of relapse was related to two variables: (1) the hospital conducting the study and (2) the dose of tranquillizing medication the patient was receiving before being put on placebo. Patients receiving low doses of tranquillizing medication before the study were less likely to relapse than patients receiving moderate to high doses. The practical and theoretical implications of these findings are discussed.