This objectives of this study were three-fold: retrospectively evaluate anxiolytic/hypnotic consumption by psychiatric inpatients, identify the risk factors of prolonged intakes, and prospectively measure the impact of hospitalisation on the use of those drugs. Three hundred and seventy-six patients hospitalised in 11 psychiatric departments in the Paris region were studied using a structured interview for the anxiolytic/hypnotic treatments, DSM-III-R criteria, GHQ-12, HAD, Spiegel's questionnaire, COVI's anxiety scale and the CGI. Eighty-five per cent of the patients had taken one anxiolytic/hypnotic or more in the 3 months preceding hospitalisation. Hospitalisation induced little change in anxiolytic/hypnotic use: dosage frequency increased from 77% to 84% between the week preceding hospitalisation and that preceding discharge; 26% of consumers were taking at least two anxiolytics or two hypnotics in the first period vs. 23% in the second. The absence of withdrawal during hospitalisation was related to the high age and a diagnosis of depression rather than schizophrenia, to the existence of continuous intake over the 3 months preceding hospitalisation and to higher drug doses during the 7 days preceding hospitalisation. Prescription of treatment at the end of hospitalisation in previously non-user subjects was related to a higher HAD anxiety score at discharge.