Abstract Background There is inconsistency in reported patient characteristics associated with differences in basal and poststress left ventricular function (delta ejection fraction [ΔEF]) assessed by gated single photon emission computed tomography (SPECT). This inconsistency may therefore hamper adequate interpretation. In this study we first determined the reproducibility of serial gated SPECT–assessed left ventricular function. Second, we determined whether left ventricular ejection fraction (LVEF) assessed directly after stress reflects basal LVEF and, if not, what patient characteristics were associated with this discrepancy in LVEF. Methods and results Serial reproducibility of technetium 99m tetrofosmin gated SPECT–assessed LVEF in 22 patients showed a mean difference between two sequential measurements at rest of 0.09% EF units, with a 95% limit of agreement (2 SDs) at 5.8% EF units. In 229 patients Tc-99m tetrofosmin gated SPECT was performed after stress and at rest. Independent predictors of ΔEF were the presence of scintigraphically proven ischemia (standardized coefficient, −1.256; P = .003) and difference in heart rate at the time of acquisition (standardized coefficient, 0.121; P = .001). Conclusions Gated SPECT–assessed LVEF at rest is reproducible under standard clinical conditions. However, LVEF assessed after stress does not represent LVEF at rest in patients with scintigraphically proven ischemia and in patients with increased heart rate after stress compared with heart rate at rest.