To evaluate the reliability of the videodensitometric assessment of right ventricular ejection fraction, 38 patients were studied during diagnostic cardiac catheterization. Digital subtraction images of the right ventricle were obtained in both the right anterior oblique and the left anterior oblique views, using direct intraventricular injection of dilute contrast medium. From the end-diastolic and end-systolic images obtained in each view, analysis of the relative brightness values generated a videodensitometry-based right ventricular ejection fraction for both the right and the left anterior oblique views. These values were compared with those generated by applying the geometry-based Simpson's rule to the orthogonal images. Right ventricular ejection fraction ranged from 22 to 88%. Videodensitometric ejection fraction in the right anterior oblique view correlated well with that in the left anterior oblique view (r = 0.88) and each correlated well with geometry-based ejection fraction (r = 0.91 and 0.82, respectively). In a subset of 18 patients without significant cardiac disease, mean videodensitometric right ventricular ejection fraction was 68% (versus 61% in the abnormal subset), and it correlated closely with left ventricular ejection fraction (r = 0.82). Videodensitometric analysis of digital subtraction images provides a reliable method for calculating right ventricular ejection fraction that is independent of geometry and reliably separates normal from abnormal values. Application of videodensitometric techniques should simplify analysis of the response of the right ventricle to different interventions in patients with cardiac disease.