OBJECTIVE To determine the spontaneous changes in volumetric indices of right ventricular function assessed by thermodilution. METHODS The study involved measurements of inter- and intraindividual variation of right ventricular function in human subjects at rest and at two different levels of low load supine bicycle exercise. METHODS The subjects were 23 patients with chronic obstructive pulmonary disease, referred for evaluation by right heart catheterisation. Mean (SEM) age was 55.4 (2.2) years and all were in sinus rhythm. RESULTS Successive measurements of ejection fraction and right ventricular end diastolic and end systolic volumes were made by thermodilution with a fast response thermistor mounted in a Swan-Ganz catheter. Five measurements were made at rest (R), and three measurements during each of two levels of exercise (E1, E2). Group average values for each period showed no significant changes with time; interindividual variability was expressed as the coefficient of variation (VC1 = SD/means), intraindividual variability (VC2) as square root of mean value of individual variances. Interindividual variability increased from rest to exercise, while intraindividual variability was lower during exercise, presumably because the variables were more steady. For ejection fraction, VC1, was 23% at rest, 29% for E1 and 32% for E2, while VC2 was 21%, 12% and 15% respectively; for right ventricular end diastolic volume, VC1 was 23% at rest, 29% and 28% during exercise, and VC2 was 17%, 12% and 11% respectively. In some patients cold injection induced bradycardia and spuriously high values of ejection fraction. CONCLUSIONS - It is important to monitor heart rate before and during thermodilution measurements and to take into account only those made with stable heart rate. Injectate temperature should be above 10 degrees C.