Abnormalities of skeletal muscle rather than of haemodynamics may be important determinants of exercise capacity in chronic heart failure. We investigated an array of indicators of central haemodynamics and peripheral muscle function to establish which resting measurements predicted exercise performance. In 20 patients quadriceps strength, resting and peak leg blood flow and leg muscle cross sectional area were measured. In 18 patients average daytime blood pressure and pulse rate, haemodynamic variables at rest and during exercise, and autonomic activity were measured. There were correlations between peak oxygen consumption and quadriceps strength (0.65; P = 0.007), thigh muscle cross sectional area (r = 0.63; P = 0.004), and average daytime systolic blood pressure (r = 0.66; P < 0.01). There were no correlations with indices of peripheral blood flow, measures of haemodynamic function, or autonomic function. Quadriceps strength was the most important individual correlate of exercise tolerance (r = 0.73). With total muscle cross sectional area and left quadriceps strength also taken into consideration, 82% of the variation in peak oxygen consumption was explained. Of the haemodynamic variables, only average daytime systolic blood pressure predicted exercise performance. The resting variables that best predict exercise performance in chronic heart failure are measures of skeletal muscle function and bulk, and average daytime systolic blood pressure. These findings suggest that abnormalities in the periphery largely determine exercise performance in chronic heart failure, and that the ability of the heart to generate an adequate blood pressure response to daily activities is also predictive of functional status.