To evaluate the additional value of transesophageal (TEE) compared with transthoracic (TTE) echocardiography and the role of patent foramen ovale (PFO) and deep vein thrombosis in the work-up of embolic events, patients with presumed cardiac embolic stroke or transient ischemic attack (neurovascular etiology was excluded) were prospectively studied by transthoracic and transesophageal contrast echocardiography. If PFO was detected echocardiographically, PFO size was assessed semiquantitatively and phlebography of both legs was performed. Two hundred forty-two consecutive patients (153 men, 60 +/- 15 years) were studied. In 197 patients, neuroimaging showed evidence of embolic infarction. TEE identified 138 potential cardiac sources of embolism in 111 patients, compared with 69 by TTE (p <0.01) in 59 patients. TEE detected potential cardiac sources in 52 patients with negative TTE examination and was significantly superior compared with TTE for identifying left atrial thrombi, spontaneous echo contrast, PFO, atrial septal aneurysm, and atheroma of the ascending aorta. In patients with a positive TTE, additional diagnostic information by TEE was found in only 6 patients and did not change therapy. Phlebography was performed in 53 patients with PFO and revealed deep vein thrombosis in 5 patients (9.5%); all had medium or large PFOs. Thus, in patients with cerebral ischemia of suspected cardiogenic origin and a normal TTE examination, TEE detects potential causes of embolism in 31% of patients and is therefore of diagnostic relevance. Conversely, in the presence of a diagnostic TTE an additional TEE confers only marginal diagnostic benefit. Deep venous thrombosis was detected in nearly 10% of patients with PFO as the sole identifiable cardiac risk factor. Given that in 4 of 5 patients deep vein thrombosis was clinically silent, phlebography should be performed in patients with medium or large interatrial shunts if paradoxical embolism is suspected.