Pathologies of cerebral circulation are one of the most frequent causes of mortality and morbidity among the populations of Western countries. The incidence of ischemic events presumed to have a cardioembolic origin varies from 13% to 34% in the largest international multi-center studies. However, some authors have reported an incidence of general cardiogenic stroke ranging from 23% to 36% in younger patients. Transesophageal echocardiography (TEE) is a useful investigation for identifying cardiac sources of embolism in patients without vascular alterations involving the carotid and vertebral districts. This study comprised 73 patients with unexplained stroke who were investigated by both transthoracic echocardiography (TTE) and TEE. Vascular echocolor sonography had been performed in all of them with negative results. TEE demonstrated: atrial or ventricular thrombosis in 6 patients (8%), intracavitary neoplasm in 2 (2%), spontaneous echocontrast in 11 (13%), valve strands in 6 (7%), complex aortic plaque in 10 (11%), patency of the foramen ovale in 17 (19%), atrial septal aneurysm in 9 (11%), dystrophy and mitral calcifications in 9 (11%). In the other 18 patients, TEE did not reveal any anomalies. Definite sources of stroke (clots and tumors) were identified in 14% of all the lesions; however, the majority (86%) were potential sources. A comparison of the two echocardiographic techniques demonstrated a greater sensitivity and specificity of TEE. Although TEE was superior to TTE for identifying and evaluating potential embolic sources, we found both methods to be useful and would advise performing TTE before TEE. While TTE is not capable of identifying the majority of anomalies, it does provide useful information to guide subsequent transesophageal investigations.