In principle, region of interest (ROI) selection should not affect the measurement of cardiac output by the first-pass technique with a radioactive intravascular indicator. Clinical application of the method requires that this theoretical hypothesis be tested. Sixty-eight left anterior oblique first-pass studies were acquired with a scintillation camera and computer using red blood cells labeled in vitro with 99mTc. Calculated mean cardiac output varied in the following order with respect to ROI: lung greater than right heart greater than left ventricle greater than whole heart (both ventricles) greater than aorta. Similar variations were observed in patients both with and without valvular regurgitation. Regions of interest over left ventricle or whole heart yielded the best correlations with cardiac output by thermodilution (r = 0.96, 0.95, respectively, n = 28) as well as the smallest interobserver variations (r = 0.994, 0.995, respectively, n = 33). First-pass studies with [99mTc]red blood cells labeled in vitro can yield accurate, reproducible determinations of cardiac output provided that the effect of ROI selection is recognized and that regions are properly selected.