The aim of the study reported here was to test the validity of a simple clinical classification of acute ischaemic stroke (Oxfordshire Community Stroke Project, OCSP) in predicting the site and size of cerebral infarction on computed tomography (CT). Consecutive patients admitted to hospital with acute ischaemic stroke were prospectively identified and classified into one of four clinical syndromes according to the OCSP classification, blind to the result of CT. The CT brain scans were classified blind to the clinical features into those demonstrating: small, medium or large cortical infarcts; small or large subcortical infarcts in the anterior circulation territory; and posterior cerebral circulation territory infarcts. A total of 108 patients were included. A recent infarct was seen on the CT scan in 91 patients (84%), and the clinical classification correctly predicted the site and size of the cerebral infarct in 80 of these (88%; 95% confidence interval 77-92%). The positive predictive value was best for large cortical infarcts (0.94) and worst for small subcortical infarcts (0.63). The OCSP clinical classification is a reasonably valid way of predicting the site and size of cerebral infarction on CT and can, therefore, be used very early after stroke onset before the infarct appears on the scan.