BACKGROUND A recent study demonstrated that in patients with nonvalvular atrial fibrillation (AF), a ratio of left ventricular ejection fraction (LVEF) to the left atrial volume index (LAVI) of <1.5 has 100% sensitivity for detecting left atrial appendage (LAA) thrombus. We sought to validate this prediction tool in an external cohort. METHODS We conducted a cohort study of consecutive AF patients who underwent transoesophageal echocardiogram (TEE) to 'rule-out' LAA thrombus and had a prior transthoracic echocardiogram (TTE). The LAVI and LVEF were measured to calculate LVEF/LAVI ratio. The sensitivity and specificity of LVEF/LAVI <1.5 were calculated. RESULTS Among 215 subjects, 19 (8.8%) had LAA thrombus and also had a higher mean CHADS2 score (2.5 vs. 1.9, P = 0.04), lower mean LVEF (24 vs. 44%, P < 0.001), higher mean LAVI (44 mL/m2 vs. 30 mL/m2, P < 0.001), and higher prevalence of cardiac failure (79 vs. 52%, P = 0.02). The LVEF and LAVI were found to be independent predictors of LAA thrombus (P < 0.05). The LVEF/LAVI ratio diagnosed LAA thrombus with an area under the curve = 0.83 by the receiver operator characteristics curve analysis (P < 0.001). All 19 (100%) subjects with LAA thrombus had LVEF/LAVI <1.5 vs. 87 (44%) among those without LAA thrombus (P < 0.001). The sensitivity and specificity of LVEF/LAVI <1.5 were 100 and 55.6%, respectively. CONCLUSIONS This investigation validates a simple TTE prediction rule to exclude the diagnosis of LAA thrombus, which may obviate the need for pre-cardioversion TEE in selected patients with nonvalvular AF.