To evaluate the relationship between cardiac function at rest and the exercise capacity of diabetic patients, left ventricular function and exercise capacity were evaluated in 15 non-insulin-dependent male diabetic patients. Isovolumic relaxation time (IRT) and the PEP/LVET ratio were obtained by simultaneous echophonocardiograms, electrocardiograms, and recordings of the carotid arterial pulse. VO2 at anaerobic threshold was obtained from a cycle ergometer exercise test with expired gas analysis. Patients were divided into two groups: those with IRT less than 90 msec (Group 1) and those with IRT greater than or equal to 90 msec (Group 2). Group 2 patients had a lower VO2 at anaerobic threshold than Group 1 (Group 1: 17.4 +/- 3.6, Group 2: 12.9 +/- 2.5 ml/min/kg; M +/- SD, p less than 0.05). There was a good correlation between the IRT at rest and delta HR, defined as the change in heart rate from rest to anaerobic threshold (r = 0.666, p less than 0.01), and between delta HR and the beat-to-beat variation in R-R interval at rest (r = 0.637, p less than 0.02). There was also a good correlation between VO2 at anaerobic threshold and IRT (r = 0.555, p less than 0.05), and between VO2 at anaerobic threshold and delta HR (r = 0.858, p less than 0.01). Our data suggests that both left ventricular diastolic function and cardiac sympathetic nervous system reflexes, reflected by delta HR, may affect the exercise capacity of diabetic patients. Thus, IRT obtained by a non-invasive method may provide a useful index for cardiovascular response to exercise in diabetic patients without coronary artery disease.