Anesthesia residents take longer to detect changes in electronically monitored data during the induction phase of anesthesia during the maintenance phase. This study was performed to investigate the reasons for this delay and to validate a method of measuring vigilance. The activity of ten residents was studied during 73 surgical procedures. Data were collected during three 15-min periods from each case: induction, starting with application of the electrocardiograph; maintenance, an arbitrary period between induction and emergence; and emergence, ending with detachment of the electrocardiograph. Vigilance was measured as the time taken to detect a change, from normal to abnormal, of an artificial parameter displayed on the physiologic monitor (response time). An observer simultaneously recorded each time that the resident looked toward the monitors. Vigilance to the monitor display was less during induction and emergence than during maintenance (P < 0.005). Residents spent less total time watching monitors during induction than during maintenance (P < 0.005), and the duration of each monitor observation was shorter (P < 0.0005). Anesthesia residents usually looked at the monitors several times before detecting the abnormal value. The measure of anesthesia vigilance correlated with independent measures of monitor watching time and frequency. The results suggest that during induction of anesthesia, which is a period of high anesthesiologist workload, residents glance toward monitors to gather data rather than scan displays. The results help to validate the method for measuring anesthesia vigilance.