This study tested the hypothesis that stroke patients without a cardiac source of embolism suspected by clinical examination can be risk stratified by transesophageal echocardiography. Forty ischemic stroke patients without atrial fibrillation, prosthetic valves, ejection fraction < 20%, or recent myocardial infarction underwent multiplane transesophageal echocardiography: 24 (designated high risk) had > or = 1 of the following: left heart thrombus, vegetation, mass or spontaneous echo contrast, mobile ascending aortic or arch debris, patent foramen ovale, atrial septal defect or aneurysm, mitral annular calcification, mitral valve thickening, prolapse or mitral valve strands. End points were death, recurrent stroke, transient ischemic attack, myocardial infarction or peripheral embolism. Thirty-eight patients (95%) (23 high, 15 low risk) were followed for 14 +/- 8 months: 9 (24%) died of vascular causes including 4 who had a cardiac cause of death and 5 who had fatal strokes. Eight had recurrent strokes (4 nonfatal) and 1 nonfatal myocardial infarction occurred. Cardiovascular survival was predicted by transesophageal echocardiography: survival rates were 92% (low risk) and 63% (high risk) at 24 months (p = 0.036). Left atrial enlargement was independently associated with death from stroke (fatal stroke occurred in 25% of those with atrial enlargement compared to 8% of those with normal atrial dimension, p < or = 0.03), as was left atrial spontaneous echo contrast (50% died vs 9% without contrast, p < or = 0.03). Left ventricular hypertrophy and aortic atherosclerosis were both associated with the risk of recurrent stroke (30% of patients with ventricular hypertrophy had recurrent stroke compared to 10% with normal wall thickness (p < or = 0.05); 30% with aortic atherosclerosis had a recurrent stroke compared to none with a normal aorta (p < or = 0.05). Thus, transesophageal echocardiography clearly identifies patients at a high risk for cardiovascular mortality and morbidity after stroke despite an unsuspected source of embolism by clinical examination.