The aim of this study was to determine whether left atrial ejection fraction (LAEF) quantified with cardiovascular magnetic resonance (CMR) was different between heart failure with preserved ejection fraction (HFpEF) and controls, and its relation to prognosis. As part of our single-centre, prospective, observational study, 188 subjects (HFpEF n = 140, controls n = 48) underwent phenotyping with contrast-enhanced CMR, transthoracic echocardiography, blood sampling and six-minute walk testing. LAEF was calculated using the biplane method. Atrial fibrillation (AF) was present in 43 (31%) of HFpEF subjects. Overall, LAEF (%) was lower in HFpEF patients inclusive of AF (32 ± 16) or those in sinus rhythm alone (41 ± 12) compared to controls (51 ± 11), p < 0.0001. LAEF correlated inversely with maximal and minimal left atrial volumes indexed (r = - 0.602, r = - 0.762), and plasma N-terminal pro-atrial natriuretic peptide (r = - 0.367); p < 0.0001. During median follow-up (1429 days), there were 67 composite events of all-cause death or hospitalization for heart failure (22 deaths, 45 HF hospitalizations) in HFpEF. Lower LAEF (below median) was associated with an increased risk of composite endpoints (Log-Rank: all p = 0.028; sinus p = 0.036). In multivariable Cox regression analysis, LAEF (adjusted hazard ratio [HR] 0.767, 95% confidence interval [CI] 0.591-0.996; p = 0.047) and indexed extracellular volume (HR 1.422, CI 1.015-1.992; p = 0.041) were the only parameters that remained significant when added to a base prognostic model comprising age, prior HF hospitalization, diastolic blood pressure, lung disease, NYHA, six-minute-walk-test-distance, haemoglobin, creatinine and B-type natriuretic peptide. CMR-derived LAEF is lower in HFpEF compared to healthy controls and is a strong prognostic biomarker.