Studies employing psychiatric rating instruments administered by lay interviewers need to demonstrate the reliability and validity of subject classification. In a national population study utilizing the short form of the Present State Examination, audiotapes of interviews were used to demonstrate the reliability of ratings. The classification of ‘cases’ was found to be related to self-reports of nervous trouble and help–seeking for such problems. Associations between the rate of psychiatric disorder and demographic factors were similar to those found in previous community surveys. In general, the method of psychiatric assessment was considered to be satisfactory. Recommendations are made for the monitoring of interviews in similar large-scale investigations.

A survey to analyse patients' characteristics and the care delivered to them was conducted in several psychiatric facilities, covering the whole range of care offered in the public sector in two regions of Southern Italy. A case–control method of analysis was chosen, and the relative risk estimates (RR), together with their 95% confidence intervals (CI), of being in custodial facilities were computed, comparing patients treated in custodial institutions (‘cases’) with those in community-based services (‘controls’) in respect of social and clinical vairables. Factors more stongly associated with custodial treatment were: illiteracy (RR = 2·4), unmarried status (RR =2·4), a history of prolonged illness (RR = 7·6), organic diagnosis (RR = 2·6), previous custodial treatment (RR = 3·9), and expected poor social functioning (RR =2·4). The suitability and advantages of the case–control method of analysis in the field of psychiatric care evaluation are discussed.