Effective strategies for the aetiologic diagnosis in patients with ischaemic stroke can be implemented based on simple clinical criteria and instrumental tests which can be performed in a modern emergency room (ER) within 24 hours from admission. This may bear prognostic and therapeutic relevance for patients with acute stroke. Therefore, in this study we set out to establish the feasibility and accuracy of the aetiologic diagnosis of ischaemic stroke in an ER. A total of 136 consecutive patients (mean age 72+/-10 years, 60 females) with first ever ischaemic stroke admitted during 1996-1997 were evaluated with assessment of clinical features, CT scan, ECG, ultrasonography of the extracranial arteries, transthoracic echocardiography, and, in selected patients, transoesophageal echocardiography. Patients were classified into two major categories defined as stroke of determined origin and stroke of undetermined origin (a stroke with two or more possible causes or with a negative evaluation), according to the TOAST criteria. Ninety-six patients were considered affected by stroke of determined origin (70.5%), (22.7% with large artery atherosclerosis, 19.1% with cardioembolic stroke, 26.4% with lacunar stroke and 1.4% with other aetiology). The remaining 40 patients (29.4%) had stroke of undetermined origin: of these, 13 patients (9.5%) had a totally negative evaluation, 15 patients (12.5%) showed cardioembolism among the two or more possible causes of stroke and seven patients (5.1%) had atherothrombotic or lacunar aetiology. Additional work-up with transoesophageal echocardiography succeeded in demonstrating aortic embolism in five patients (3.6%; i.e. four patients with aortic plaques more than 4 mm in thickness and one patient with ulcerated plaques). In conclusion, the subtype classification system for ischaemic stroke allowed the aetiological diagnosis in 70.5% of patients while in the ER. Stroke of undetermined origin represented one-third of patients in a consecutive population with acute onset neurologic deficit of ischaemic origin. In approximately half of the patients with negative standard evaluation, cardiogenic or aortic arch embolic sources could be identified by transoesophageal echocardiography. Thus, the latter is indicated in patients with stroke of undetermined origin with negative first-line evaluation in order to identify embolic sources in the aortic arch.