Although exercise intolerance is a cardinal symptom of patients with dilated cardiomyopathy (DC) and heart failure, the factors that limit exercise capacity in these patients remain a matter of debate. To assess the contribution of left ventricular (LV) diastolic filling to the variable exercise capacity of patients with DC, we studied 47 patients (60 +/- 12 years) with DC in stable mild-to-moderate heart failure with a mean LV ejection fraction of 28%. Exercise capacity was measured as total body peak oxygen consumption (VO2) during symptom-limited bicycle (10 W/min) and treadmill (modified Bruce protocol) exercise. LV systolic function and diastolic filling were assessed at rest before each exercise by M-mode, Doppler echocardiography, and radionuclide ventriculography. As expected, treadmill exercise always yielded higher peak VO2 than bicycle exercise (21 +/- 6 vs 18 +/- 5 ml/kg/min, range 12 to 35 and 7 to 30 ml/kg/min, respectively, p <0.001). Both of these VO2 measurements were highly reproducible (R = 0.98). With univariate analysis, close correlations were found between peak VO2 (with either exercise modalities) and Doppler indexes of LV diastolic filling, as well as with the radionuclide LV ejection fraction. Stepwise multiple regression analysis identified 3 nonexercise variables as independent correlates of peak VO2, of which the most powerful was the E/A ratio (multiple r2 = 0.38, p <0.0001), followed by peak A velocity (r2 = 0.54, p <0.0001) and mitral regurgitation grade (r2 = 0.58, p = 0.024). In conclusion, our data indicate that in patients with DC, peak VO2 is better correlated to diastolic filling rather than systolic LV function.