In stroke patients several cardiac changes associated with embolism can be detected with transoesophageal echocardiography. Potential major cardiac embolic sources (e.g. atrial fibrillation, thrombi of left ventricle/atrium, vegetation, myxoma, dilated cardiomyopathy) have a causal relationship to embolism. Other changes with no certain causal relationship are regarded as potential minor cardiac embolic sources (e.g. atrial septal aneurysm, patent foramen ovale, mitral annular calcification, mitral valve prolapse, protruding atheroma of the aorta). We compared the prevalences of major and minor potential cardiac embolic sources in a stroke population with that in controls. One hundred and twenty-one patients with first-ever stroke were compared with 68 randomly selected controls. All subjects underwent magnetic resonance imaging of the brain, carotid ultrasound and transthoracic/transoesophageal echocardiography. The patients were slightly older (mean age 70.7 +/- 10.3 years) than the controls (65.5 +/- 15.5 years) (p < 0.05). Potential major cardiac embolic sources were found in 27% of the patients and in 4% of the controls (p < 0.001). The most common major potential embolic source was atrial fibrillation, detected in 22/121 patients. Fifteen of these also had spontaneous echocontrast in the left atrium. Eleven left atrial thrombi were found (four of these patients had atrial fibrillation and seven had sinus rhythm). A history of heart disease was more common in patients with a potential major cardiac embolic source or a carotid artery stenosis (77%) than in those patients without (44%) (p < 0.01). After excluding subjects with a major potential cardiac embolic source and/or carotid artery stenosis, no differences in the prevalence of minor potential cardiac embolic sources were found between patients (55%) and control subjects (47%) (p = NS). Even when subjects without a major potential cardiac embolic source or a carotid artery stenosis were categorized into three age groups (35-54, 55-74 and > 74 years) the prevalence of potential minor cardiac embolic sources did not differ between patients and controls. To conclude, major potential cardiac embolic sources are more common in an older population with first-ever stroke than in a comparable control group. However, potential minor cardiac embolic sources did not differ in prevalence in the patients compared with controls. Certain changes (e.g. atrial septal aneurysm) might have a potential embolic role in younger stroke patients but in our study no difference was found between older stroke patients and controls.