The relation between bicycle exercise performance and determinants of central and peripheral cardiovascular function was assessed in 17 patients with class II to III heart failure and in 9 normal subjects. Proportional changes in oxygen consumption (VO2) from rest (R) to maximal exercise (Ex), or Mets, were used as an objective measure of the exercise capacity or functional reserve of the cardiovascular system. Left ventricular ejection fraction (EF) and proportional changes in end-diastolic volume, stroke volume, and cardiac output were determined from appropriate count data by equilibrium radionuclide angiography. Proportional changes in arteriovenous oxygen difference (A-VO2) were derived from the equation Ex/R A-VO2 = Ex/R VO2 divided by Ex/R CO, where CO = count output. Each subject exercised to an anaerobic endpoint. Maximal VO2 was significantly lower in patients than in normal subjects. Because Ex/R A-VO2 was comparable in normal subjects and patients, the lower exercise performance in patients resulted from a reduced count output response. The reduced CO response in patients resulted from failure of the ejection fraction to increase or from an attenuated heart rate response, or both. Exercise performance was variable in both groups. Multivariable analysis in the patient group identified changes in heart rate, count output, and A-VO2 with exercise as important predictors of Mets, but found no relation between Mets and changes in ejection fraction or stroke counts during exercise. Similarly, multiple regression analyses between Mets and determinants of cardiovascular function demonstrated significant correlations with Ex/R heart rate, Ex/R count output, and Ex/R A-VO2 in both groups. In patients, EF at rest ranged from 0.09 to 0.36, but it did not correlate with Mets, nor did changes in ejection fraction, stroke counts, or end-diastolic counts during exercise. The variable exercise performance among patients with severe left ventricular dysfunction was determined predominantly by a variable heart rate and A-VO2 response and not by rest or exercise indexes of left ventricular function.