Since 1961 the Scottish Home and Health Depart ment has collected and processed morbidity infor mation relating to all non-obstetric and non-psy chiatric discharges from National Health Service hospitals in Scotland. The information is used for planning, administrative, and epidemiological pur poses and it is therefore essential that its reliability is assessed. The present study was carried out to obtain factual information relating to the accuracy of the data collected, no previous studies having been undertaken in Scotland to evaluate the national hospital morbidity information. Morbidity information concerning each hospital discharge is recorded on the form S.M.R.I, usually by hospital medical records staff. This basic data collection sheet, the checking and elimination of errors, and the uses to which the data are put have been fully described by Heasman (1968). These data are the basis of the Scottish Hospital In-patient Statistics published annually, the unpub lished tabulations sent to all hospitals and to regional hospital boards, the individual hospital diagnostic and operation indexes, and the recently introduced Scottish Consultants Review of In-patient Statistics. Each form received in the central department is scanned for obvious errors and omissions and is then subjected to a computer data vetting and feasibility check. These checks attempt to ensure that as many inconsistencies as possible are detected, and enable any necessary corrections to be made before final acceptance. It is, however, possible for errors to remain undetected if the information, although inaccurate, is neither infeasible nor inadmissible. The present paper reports on the accuracy of the information received for processing in the year 1969. Errors in transcription and coding of the information recorded in the patient's medical case record are described, and an attempt is made to assess the relative importance of the levels of error found.