ABSTRACT Background Approximately 20% of cerebral infarctions are caused by a cardiac source of embolism. Although transesophageal echocardiography (TEE) is a valuable tool for evaluating patients with suspected cardioembolism, its use should be more selective. Objective We attempted to determine whether risk factors derived from the clinical, ECG, and transthoracic echocardiogram (TTE) would make TEE use more selective and increase its yield. Methods We retrospectively reviewed the records of 108 patients with suspected embolic stroke who had undergone TEE evaluation. We defined the presence of left atrial appendage spontaneous contrast (LAASC), left atrial appendage thrombus (LAAT), patent foramen ovale (PFO), or intraatrial septal aneurysm (IASA) as TEE endpoints suggestive of cardioembolism. We evaluated the association between the TEE endpoints and (1) age greater than 60 years, (2) the presence of atrial flutter (AFl) or atrial fibrillation (AF) on ECG, (3) left ventricular (LV) dysfunction by TTE (mild, LVEF 4 cm also determined by TTE. The statistical analysis performed was one-way analysis of variance controlling for interactions between different risk factors and the endpoints. Results The strongest independent predictors for the presence of LAASC or LAAT were age greater than 60 years, presence of AFl/AF, and LV systolic dysfunction. There was no association between any of the risk factors with the presence of PFO or IASA. Conclusion This study demonstrates that clinical characteristics, ECG, and TTE findings can help to better select stroke patients for TEE. Future larger studies are needed to provide more supporting data.