Visual monitoring at the central station of coronary care units (CCUs) may not adequately assess the presence and frequency of transient asymptomatic ST segment changes in patients with unstable angina. We have performed continuous 24-h electrocardiographic recordings over a total period of 50 days in 10 patients admitted to our CCU with frequent attacks of angina at rest. Over the corresponding period, at the central monitoring station (6 beds), the nursing-staff detected only 31 transient ischaemic episodes (27 with ST elevation, 4 with depression, 9 of which were asymptomatic). By contrast the retrospective analogue analysis of tapes identified 213 ischaemic episodes: 143 with ST elevation (greater than or equal to 0.2 mV) and 70 with ST depression (greater than or equal to 0.2 mV); usual CCU monitoring failed to report changes during 13 episodes (8 with ST elevation, 5 with depression) accompanied by anginal pain. When the tapes were played back in real time on a CCU monitoring scope, a cardiologist (who had the option of interrupting the play-back whenever tired) recognized 48% of the episodes when presented in groups of 4 and randomly positioned on the screen together with 2 other electrocardiographic tracings not related to the study; he recognized 92% of the episodes when only one ECG was presented on the screen. Thus conventional visual monitoring in a CCU considerably underestimates the incidence of transient ischaemic ST segment changes, some of which were accompanied by pain. This low rate of detection is the result of the presentation on the central monitoring station of several ECGs and of fatigue.