The demands of dynamic monitoring and fault diagnosis for flight engineer trainees were examined in relation to changes in heart rate (HR) and two spectral analysis measures (midfrequency: 0.07-0.14 Hz; high frequency: 0.15-0.40 Hz) of heart rate variability (HRV). Eleven trainee flight engineers were studied, as part of their training and assessment, over three 3-h sessions in a cockpit simulator. During each session, faults and incidents programmed into the system had to be detected, diagnosed, and corrected. Electrocardiograms were taken, and each session was recorded on videotape. Work phases were classified from video analysis of flight maintenance activities, using Rasmussen's cognitive control taxonomy, into monitoring, routine (rule-based), and problem-solving (knowledge-based) phases. HR and HRV were found to be sensitive to different phases of the work environment. HRV was suppressed during the mentally demanding problem-solving mode of the level flight phase, but only for the midfrequency component. Elevated heart rate, in contrast, was associated with the more generally stressful takeoff and landing phases. The findings support both the use of HRV as a physiological index of mental effort and its value in operational contexts, and the value of ecologically derived methods of evaluating differences in work demands in complex systems.