Abstract Background For the primary prevention of sudden cardiac death, guidelines provide left ventricular ejection fraction (EF) criteria for implantable cardioverter defibrillator (ICD) placement without specifying the technique by which it should be measured. We sought to investigate the potential impact of performing cardiovascular magnetic resonance (CMR) for EF on ICD eligibility. Methods The study population consisted of patients being considered for ICD implantation who were referred for EF assessment by CMR. Patients who underwent CMR within 30 days of echocardiography were included. Echocardiographic EF was determined by Simpson’s biplane method and CMR EF was measured by Simpson’s summation of discs method. Results Fifty-two patients (age 62±15 years, 81% male) had a mean EF of 38 ± 14% by echocardiography and 35 ± 14% by CMR. CMR had greater reproducibility than echocardiography for both intra-observer (ICC, 0.98 vs 0.94) and inter-observer comparisons (ICC 0.99 vs 0.93). The limits of agreement comparing CMR and echocardiographic EF were – 16 to +10 percentage points. CMR resulted in 11 of 52 (21%) and 5 of 52 (10%) of patients being reclassified regarding ICD eligibility at the EF thresholds of 35 and 30% respectively. Among patients with an echocardiographic EF of between 25 and 40%, 9 of 22 (41%) were reclassified by CMR at either the 35 or 30% threshold. Echocardiography identified only 1 of the 6 patients with left ventricular thrombus noted incidentally on CMR. Conclusions CMR resulted in 21% of patients being reclassified regarding ICD eligibility when strict EF criteria were used. In addition, CMR detected unexpected left ventricular thrombus in almost 10% of patients. Our findings suggest that the use of CMR for EF assessment may have a substantial impact on management in patients being considered for ICD implantation.