Within the setting of a regional secure unit, all doses of medication given p.r.n. over three months were ascertained and the details of each administration determined from prescription charts and a semistructured interview with the nursing staff involved. Thirty-two patients were resident for all or part of the study, all compulsorily detained. Only 15 were given medication p.r.n.; compared with the other patients they were significantly younger and more likely to be detained under civil orders than under the criminal provisions of the Mental Health Act 1983. Rarely was the medication the sole intervention in a crisis. In terms of effects on the overall treatment programme of the patient, medication given p.r.n. seemed to have an impact in only one small subgroup.