We reviewed 39 medical audits started in one institution between 1981 and 1985 to determine whether they monitored the quality of health care in a scientific manner, whether recommendations were made that could improve the quality of health care and, if so, whether the recommendations were acted on. Thirty-three audits (85%) were completed; 21 (64%) failed to state their objectives, 30 (91%) failed to compare their results with those in the literature, 9 (27%) made no recommendations that could improve the quality of health care, and 9 were poorly researched and written. The number of patients in each audit was usually adequate. Feedback was rarely received by the Medical Audit Committee concerning the Medical Advisory Committee's response to the audit. Information concerning the implementation of recommendations was available for 17 of the 24 audits that made recommendations; 7 (41%) had failed to implement half or more of the recommendations. Despite these problems, 8 (24%) of the audits were considered to be of a quality that could improve health care. We present recommendations to improve the audit procedure and foster the growing confidence among doctors in the medical audit.