Ejection fraction and end-diastolic volume (EDV) are valuable parameters of left ventricular function. They can be measured by radionuclide ventriculography. This study examines EDV in comparison to two other methods. A comparison was made between EDV derived from radionuclide ventriculography (RNV), standard contrast ventriculography and a 'hybrid' method, which calculates EDV from ejection fraction, heart rate and thermodilution cardiac output. Two groups of patients were examined: 54 patients with various cardiac diseases by contrast and radionuclide ventriculography (RNV); and 18 patients with typical angina were examined by RNV at rest and during exercise three times: Control without medication, 20 minutes after 10 mg metoprolol i.v. and again 10 minutes after the second exercise and 10 mg isosorbide dinitrate sublingually. In comparison to contrast ventriculography RNV overestimated EDV by 27%, and in comparison to the hybrid method by 11% with satisfactory linearity. The extent of error increases with decrease of ventricular volume. The assumption that the two short axes of the left ventricle are equally long causes one quarter of total error. Errors caused by the algorithm delineating the borders of the left ventricle cannot be measured. For clinical purposes radionuclide ventricular EDV seems to yield satisfactory results.