We aimed to study the potential influence of the variability in the assessment of echocardiographically measured left ventricular ejection fraction (LVEF) on indications for the implantation of internal cardioverter defibrillator and/or cardiac resynchronization devices in heart failure patients. TIME-CHF was a multicenter trial comparing NT-BNP versus symptom-guided therapy in patients aged ≥60 years. Patients had their LVEF assessed at the recruiting centre using visual assessment, the area-length or biplane Simpson's method. Echocardiographic data were transferred to the study core-lab for re-assessment. Re-assessment in the core-lab was done with biplane Simpson's method, and included an appraisal of image quality. 413 patients had the LVEF analyzed at the recruiting centre and at the core lab. Image quality was optimal in 191 and suboptimal in 222. Overall, the correlation between LVEF at the recruiting centres and at the core-lab was good, independent of image quality (R² = 0.62). However, when a LVEF ≤30 % or ≥30 % was used as a cut-off, about 20 % of all patients would have been re-assigned to having either a LVEF above or below the cut-off, this proportion was not significantly influenced by image quality. We conclude that correlation between LVEF assessed by different centres based on the same ultrasound data is good, regardless of image quality. However, one fifth of patients would have been re-assigned to a different category when using the clinically important cut-off of 30 %.