Because reporting of complications by physicians is not reliable, monitoring of the hospital course of surgical patients by nurses can improve effectiveness of identifying complications. Monitoring the course of only those patients who remain in the hospital more than four postoperative days will disclose 90 per cent of complications expected in all patients undergoing operation. Twenty hours per week of a nurse's time is sufficient to monitor 1,300 such patients annually. Identification of patients with large hospital bills will include most patients with complicated hospital courses but does not discriminate them from the large number of uncomplicated courses which also generate large hospital charges.