Background: A reduced arterial compliance increases systolic blood pressure (BP) and consequently enhance left ventricular (LV) afterload. This phenomenon might to be exaggerated during exercise. We hypothesized that arterial compliance assessed during exercise would be more predictive for exercise capacity than that at rest. To prove our hypothesis, we measured pulse wave velocities (PWV) and central pressure wave reflection at rest as well as at immediately after peak exercise. Methods: In 91 subjects (32 men, 58짹11 years) who were referred to stress echocardiography laboratory, multistage (25W, 3 minutes increments) supine bicycle exercise testing and radial artery tonometry (SphygmoCor짰, AtCor Medical) were simultaneous performed. The patients who had significant arrhythmia, valvular disease, coronary artery disease and renal insufficiency were excluded. Central BP and augmentation index (AIx) were obtained at rest as well as immediately (< 1 min) after peak exercise using radial artery tonometry. Results: Mean exercise duration was 561 짹 189 sec. In simple correlation analysis, age (r=-0.438, p <0.001), PWV at rest (r=-0.412, p <0.001), resting AIx (r=-0.437, p<0.001), AIx after peak exercise (r=-0.525, p<0.001) and ��AIx (from rest to peak exercise) (r=-0.241, p=0.021) were significantly correlated with exercise duration. Multiple linear regression analysis, controlled for variables such as age, gender, LV mass index, revealed that AIx after peak exercise and ��AIx were independent predictors for exercise duration (棺 = -0.264, p=0.006, 棺 = -0.198, p=0.006) whereas there were no independent correlations of resting AIx and PWV with exercise duration (棺 = 0.123, p=0.201, 棺 = 0.082, 0.27) Conclusions: Unlike AIx and PWV measured at rest, AIx immediately after peak exercise and change of AIx from rest to exercise were independent predictors of execise capacity.