BACKGROUND: Previous work suggesting a better correlation of diastolic than systolic function with exercise capacity in heart failure may reflect the relative insensitivity and load-dependence of ejection fraction (EF). We sought the correlation of new and more sensitive methods of quantifying systolic and diastolic function and filling pressure with functional capacity. METHODS: We studied 155 consecutive exercise tests on 95 patients with congestive heart failure (81 male, aged 62 +/- 10 years), who underwent resting 2-dimensional echocardiography and tissue Doppler imaging before and after measurement of maximum oxygen uptake (peak VO2). RESULTS: The resting EF was 31% +/- 10% and a peak VO2 was 13 +/- 5 mL/kg.min; the majority of these patients (80%) had an ischemic cardiomyopathy. Resting EF (r = 0.14, P = .09) correlated poorly with peak VO2 and mean systolic (r = 0.23, P = .004) and diastolic tissue velocities (r = 0.18, P = .02). Peak EF was weakly correlated with the mean systolic (r = 0.18, P = .02) and diastolic velocities (r = 0.16, P < .04). The mean sum of systolic and diastolic velocities in both annuli (r = 0.30, P < .001) and E/Ea ratio (r = -0.31, P < .001) were better correlated with peak VO2 . Prediction of peak VO2 was similar with models based on models of filling pressure (R = 0.61), systolic factors (R = 0.63), and diastolic factors (R = 0.59), although a composite model of filling pressure, systolic and diastolic function was a superior predictor of peak VO2 (R = 0.69; all P < .001). CONCLUSIONS: The reported association of diastolic rather than systolic function with functional capacity may have reflected the limitations of EF. Functional capacity appears related not only to diastolic function, but also to systolic function and filling pressure, and is most closely associated with a combination of these factors.