Many patients have clinical, structural or bio-marker evidence of heart failure (HF) but a normal left ventricular ejection fraction (LVEF; HeFNEF). Measurement of global longitudinal strain (GLS) may add diagnostic and prognostic information. Patients with symptoms suggesting heart failure and LVEF ≥50% were studied: 76 had no substantial cardiac dysfunction (left atrial diameter (LAD) <40 mm and amino-terminal pro-brain natriuretic peptide (NTproBNP) <400 ng/l); 99 had "possible HeFNEF" (LAD ≥40 mm or NTproBNP ≥400 ng/l); and 138 had "definite HeFNEF" (LAD ≥40 mm and NTproBNP ≥400 ng/L). Mean LVEF was 58% in each subgroup. Patients with definite HeFNEF were older, more likely to have atrial fibrillation, had more symptoms and signs of fluid retention, were more likely to have right ventricular dysfunction and had higher pulmonary pressures than other groups. Mean GLS (SD) was less negative in patients with definite HeFNEF (-13.6 (3.0)% vs. possible HeFNEF: -15.2 (3.1)% vs. no substantial cardiac dysfunction: -15.9 (2.4)%; p < 0.001). GLS was -19.1 (2.1)% in 20 controls. During a median follow up of 647 days, cardiovascular death or an unplanned hospitalisation for heart failure occurred in 62 patients. In univariable analysis, GLS but not LVEF predicted events. However, in a multi-variable analysis, only urea, NTproBNP, left atrial volume, inferior vena cava diameter and atrial fibrillation independently predicted adverse outcome. GLS is abnormal in patients who have other evidence of HeFNEF, is associated with a worse prognosis in this population but is not a powerful independent predictor of outcome.