The aim of this study was to investigate the use of an automated three-dimensional myocardial contour detection method using elastic surfaces for the assessment of left ventricular ejection fraction (EF) from electrocardiographically (ECG) triggered myocardial perfusion single-photon emission tomography (SPET). The validity of this method was studied on the basis of both phantom measurements and patient studies. Phantom measurements were performed using an elastic phantom of the left ventricle simulating a beating heart, with a simulated EF ranging from 10% to 78%. The data from 27 patients who had undergone both ECG-triggered myocardial perfusion SPET and planar gated radionuclide ventriculography (RNV) were used to compare the EF derived from the SPET data with the automated contour detection method and the EF derived from the RNV data with standard analysis software. EF values as measured by RNV ranged from 11% to 68%. The overall accuracy of the automated contour detection method proved to be very high. In the phantom study the deviation of the measured EF from the reference values was less than 4% for all of the simulated EFs. The studies on the patient data yielded a correlation coefficient (Pearson) greater than 0.94 as compared with planar RNV. Intra- and inter-observer reproducibility was high, with correlation coefficients exceeding 0.97. It is concluded that the proposed method allows accurate, reproducible and fast measurement of the left ventricular EF on the basis of myocardial perfusion SPET.