This study assessed the prognostic value of peak cardiac power output, measured non-invasively during maximal cardiopulmonary exercise testing, against other exercise-derived haemodynamic variables in patients with chronic congestive heart failure. Two hundred and nineteen unselected, consecutive patients with congestive heart failure (166 men, mean (+/-SD) age of 56+/-13 years) who underwent maximal symptom limited cardiopulmonary treadmill exercise testing with non-invasive estimation of cardiac output using carbon dioxide re-breathing techniques, were followed-up for a mean period of 4.64 (4.47--4.82, 95% CI) years. Cardiac power output was calculated from the product of cardiac output and mean arterial blood pressure. All cause mortality was 12.3% (27 deaths). Peak and resting cardiac power output, peak mean arterial blood pressure, peak and resting cardiac output and peak VO(2)were all predictive of outcome on univariate analyses. Peak cardiac power output, either entered continuously or categorically with a cut-off value of 1.96 watts, was the only independent predictor of mortality (P=0.0004 for values < or >1.96 watts and P=0.001 for continuous values) using multivariate analysis. A relative risk ratio of 5.08 (1.94-13.3, 95% CI) was obtained for a cardiac power output <1.96 watts. Peak cardiac power output is an independent predictor of mortality that can be measured non-invasively using cardiopulmonary exercise testing. It can give further prognostic power to a peak VO(2)in the assessment of patients with congestive heart failure. Copyright 2001 The European Society of Cardiology.