Methods for angiographic determination of right ventricular volume were compared in right ventricular cast studies and in vivo using a semiautomated Simpson's rule, elliptical and rectangular cross sections and a simple geometric model, a prism with a triangular base. In cast studies excellent correlation coefficients were obtained for all three methods ( r = 0.97 to 0.98). In vivo, excellent linear correlation was obtained when the methods were compared ( r = 0.98 to 0.99, P < 0.001). Normal mean right ventricular volumes (70 patients) in milliliters per square meter±1 standard deviation were: end-diastolic volume 64 ± 15, end-systolic volume 25 ± 8 and stroke index 39 ± 11; ejection fraction was 0.61 ± 0.08. Normal mean left ventricular end-diastolic and end-systolic volumes (77 patients) were lower (59 and 19, respectively), and ejection fraction was higher (0.68). Stroke index was the same. Other results showed decreased mean end-diastollc and end-systolic volumes and stroke index, increased ejection fraction ( P < 0.001, < 0.025, < 0.025 and < 0.005, respectively) in 34 patients with pulmonary stenosis, and increased mean end-diastolic and end-systolic volumes and stroke index ( P <0.001, <0.050, <0.001, respectively) in 11 patients with an atrial left to right shunt. In both groups postoperatively, as well as in patients with a ventricular septal defect, tetralogy of Fallot or myocardial disease, mean volumes and ejection fraction were normal. We conclude that a geometric model can be used to determine right ventricular volume with accuracy. In children, the right ventricle can adapt reversibly to chronic increases in afterload and preload; it is not necessarily overloaded by a ventricular septal defect or tetralogy of Fallot, and it is relatively unaffected by severe myocardial disease.