Transesophageal echocardiography (TEE) is considered a basic tool in the diagnostic and follow-up evaluation of stroke patients, since up to 40% of cerebral ischemic events are presumed to have a cardiac origin. TEE offers a superior resolution of the posterior cardiac structures, such as left atrium and appendage and atrial septum, as well as of the aorta. By means of TEE, evidence has accumulated that some cardiovascular abnormalities (left-sided thrombi, tumors and vegetative lesions, complicated plaques of the aortic arch) are associated with ischemic stroke. Nevertheless, some issues remain unresolved. Will exclusion of atrial thrombus by multiplane TEE preclude embolism after cardioversion of atrial fibrillation? If anticoagulation before and after cardioversion is needed to provide adequate protection against embolism, will TEE be indicated in all patients? Moreover, can the detection of spontaneous echo contrast or enlarged and hypokinetic left atrial appendage in atrial fibrillation modify the therapeutic strategy? Is atrial septal aneurysm (ASA) a real embolic source, particularly when a right-to-left shunt is not associated? Considering the high prevalence of patent foramen ovale (PFO) in normal subjects, how can we identify patients at higher risk of embolism? Furthermore, methodologic points have to be taken into account when we analyze data from the literature. First, most studies are retrospective; a sole prospective study demonstrated that atherosclerotic plaques >4 mm thick in the aortic arch are significant predictors of recurrent brain infarction and other cardiovascular events in patients > or =60 years of age. Second, the association between the aforementioned cardiac abnormalities (mainly ASA and PFO) and cardiogenic embolism is biased by the patient-enrollment criteria used in those studies so that their pathogenetic role has not yet been established. Prospective studies with the enrollment of appropriate control groups will be necessary to define what can be considered a marker of embolic risk; the diagnosis "cardiogenic embolism" will not be a definitive diagnosis in most cases.