To assess the reliability of right ventricular ejection fraction (RVEF) during exercise measured by thermodilution technique using a modified Swan-Gantz catheter with a fast-response thermister, we measured RVEF under several conditions in 19 patients with cardiac disease. Measurements were repeated 5 times in each condition, and average RVEF and coefficient of variation (CV) were evaluated. 1) Injectate volume did not affect RVEF and CV. 2) A reduction in RVEF occurred with the thermister moved from proximal portion to distal portion within the pulmonary artery. 3) There were no differences in measurements of RVEF and CV between those during spontaneous breathing and those during apnea. 4) Postural change from supine to sitting decreased RVEF (38 +/- 8 to 35 +/- 9%; p < 0.05) and increased CV (7 +/- 2 to 13 +/- 5%). 5) Exercise increased RVEF (35 +/- 9 to 37 +/- 10%; p < 0.05) but did not change CV (13 +/- 5 vs 13 +/- 5%) compared with rest in the sitting position. 6) Cardiac rhythm (sinus vs atrial fibrillation) did not affect CV. 7) Average value of RVEF and CV during exercise were not different among 3, 4, 5 times repeated measurements. We considered that thermodilution technique for RVEF was applicable to exercise test, and 3 measurements were enough to determine the average value of RVEF during exercise.