The reproducibility of blood gas exchange measurements on exercise in chronic cardiac failure has already been established in patients familiar with this technique. The aim of this prospective study was to evaluate the reproducibility of cardiopulmonary parameters on exercise in a population of patients who had never undergone this type of investigation. Twenty patients with chronic cardiac failure in classes I to III of the NYHA classification, with a mean age of 55 +/- 11.5 years and a mean LV ejection fraction of 31.2 +/- 9%, underwent two cardiopulmonary exercise tests (CPX Medgraphic) performed on a bicycle ergometer. Patients underwent maximal exercise stress testing attaining 89% of the theoretical maximal heart rate and 1.14 of the respiratory quotient during the first test. There was no significant change in peak VO2 (22.5 ml/min/kg vs 22.6 ml/min/kg) or in ventilatory anaerobic threshold (12.8 ml/min/kg vs 12.7 ml/min/kg) between the two tests. The ventilatory anaerobic threshold could not be measured in one patient and seemed less reproducible than peak VO2 with a standard deviation of relative differences (T2-T1/T1) of 10.4 versus 7.8%. There was a significant increase in the duration of exercise (7.4 +/- 9.2%; p < 0.002) and a ventilatory flow (4.5 +/-, p = 0.03). This study shows that peak VO2 is a reproducible measurement in mild to moderate chronic cardiac failure, even in the absence of a preliminary test to familiarize the patient with the equipment. The reproducibility of the ventricular anaerobic threshold is less satisfactory than that of peak VO2. The increase in the duration of exercise is more dependent on motivation and should not be taken into account alone in the functional evaluation of chronic cardiac failure.