Count rates observed at end-diastole and end-systole in the region of the left ventricle on gated radionuclide angiograms were correlated with corresponding estimates by biplane contrast angiography of left ventricular volumes in 37 patients. When count rates were normalized for number of cardiac cycles processed, time per frame, isotope decay, and count rate observed in venous blood after equilibrium, there was a strong correlation between normalized left ventricular count rate and contrast angiographic volume, whether the venous sample was counted in a well counter or with the use of the same gamma camera. The latter technique was judged easier and more accurate. The derived regression equations yielded a good estimate of ventricular volume from left ventricular count rates when prospectively compared with contrast angiographic volumes in a further 17 patients. Interobserver variability of both the radionuclide and the contrast angiographic estimate of the left ventricular volume was similar. Changes in left ventricular stroke volume induced by sublingual administration of nitroglycerin as measured by the radionuclide method correlated reasonably with corresponding changes in stroke volume measured by the dye-dilution method. It is concluded that left ventricular volumes obtained noninvasively by the radionuclide technique were accurate and were subject to the same interobserver variability as the contrast angiographic technique.