The demonstration of a cardiac source of systemic embolism in patients who have suffered a cerebral ischemic event may have important therapeutic implications. This explains the large demand for echocardiography and Holter monitoring in these patients. The frequency of cerebral embolism of cardiac origin, the simplification of the diagnostic approach by non-invasive investigations and the precision of ultrasound techniques explains the tendency towards the indiscriminate generalisation of this attitude. However, the large number of potential patients for investigation, the limited facilities of investigation and the incertitude over the responsibility of certain cardiac abnormalities with respect to the context and age, are arguments in favour of a more selective investigative approach. The keystone of diagnosis is careful history taking and clinical examination with interpretation of the ECG and chest X-ray. Three clinical situations may then be identified: 1) A cardiac abnormality known to be highly embolic is diagnosed from the outset (e.g. mitral stenosis, valve prosthesis, endocarditis, myocardial infarction). The diagnostic work-up is no longer etiological: echocardiography may show intracardiac thrombi or a valvular vegetation, reinforcing the causal relationship, but the complementary investigations are mainly useful for evaluation the cardiac disease and for deciding on curative or preventive therapy. 2) A cardiac abnormality is diagnosed but its responsibility is doubtful due to its high prevalence and low embolic potential. This is the case of patients with mitral valve prolapse, mitral annular calcification, calcific aortic stenosis and VVI pacing. Complementary investigations are not discriminative for the etiological diagnosis of the cerebral embolism.(ABSTRACT TRUNCATED AT 250 WORDS)