A commercial ECG gate was tested to evaluate its ability to predict accurately the time of end-systole. The predicted times followed the manufacturer's specifications quite well. These times were compared with the actual times of end-systole as determined by computer-derived left-ventricular time-activity curves using Tc-99m-labeled red blood cells. Although there was moderate scatter, the predicted times of end-systole correlated well with the actual times (n = 59, r = 0.829). If the left-ventricular ejection fraction was calculated using the predicted time of end-systole, the error would be 0.03, or less, for 95% of the subjects.