Left ventricular (LV) systolic dysfunction, as indicated by a reduced LV ejection fraction (EF) is a potent predictor of cardiovascular mortality. Radionuclide angiography accurately and reproducibly assesses LVEF; however, echocardiography is used more frequently in clinical practice. Whether these methods predict similar mortality has not been fully investigated. We performed a retrospective analysis of patients with baseline radionuclide angiographic (RNA; n = 4,330) and echocardiographic (echo; n = 1,376) based EFs ≤0.35 who were enrolled in the Studies Of Left Ventricular Dysfunction (SOLVD) to address this hypothesis. After adjusting for important prognostic variables, the risk of death (RR 1.15; 95% confidence interval 1.01 to 1.30; p = 0.03) and of cardiovascular death (RR 1.15; 95% confidence interval 1.01 to 1.32; p = 0.04) was higher for patients with ECG-based EFs. To compare the 2 techniques across a range of EF values, we divided the cohort into tertiles of EF. The adjusted risk estimates for all-cause and cardiovascular mortality were similar within each tertile. Of note, the mortality difference in patients with echo- versus RNA-based EFs was most prominent in women. Further, patients with echo-based EFs had significantly higher mortality at sites where this technique was less frequently used to assess the EF. Thus, for a given EF ≤0.35, an echo-based value was associated with a higher risk of death compared with the RNA-based method of measurement. These data suggest that EF values determined by echocardiography and radionuclide angiography predict different mortality and this may, in part, be related to technical proficiency as well as patient characteristics.