Background: the Oxfordshire Community Stroke Project (OCSP) devised a simple clinical classification for acute stroke which predicted mortality, functional recovery and patterns of recurrent stroke. We aimed to determine whether this could predict the presence of carotid disease and be used to select which patients with acute stroke should be referred for carotid imaging with a view to subsequent carotid endarterectomy. Methods: we assessed patients with acute stroke admitted to seven hospitals over a 10-month period. Patients were classified according to the OCSP system and their carotid arteries investigated using portable continuous-wave Doppler. Those with abnormal portable assessments had colour duplex Doppler imaging. Results: of 305 patients with proven or probable cerebral infarction, severe (70–99%) ipsilateral carotid stenosis was found in 16 (16%) of the 101 with partial anterior circulation infarct (PACI), four (4%) of the 100 with total anterior circulation infarct (TACI), none of the 80 with lacunar infarct (LACI) and one (4%) of the 24 with posterior circulation infarct (POCI). Complete ipsilateral carotid occlusion was found in 25 (25%) of the TACI group, 11 (11%) of the PACI group, three (4%) of the LACI group and none of the POCI group. Severe carotid stenosis or occlusion was more common in the ipsilateral than the contralateral carotid artery for the TACI and PACI groups (x2 P<0.05), but there was no difference between ipsilateral and contralateral carotid disease in the LACI and POCI groups. If the OCSP classification is used to detect patients with 70–99% carotid stenosis, then the sensitivity is 76% and specificity is 70%. Conclusion: these findings suggest that ipsilateral carotid disease is an important cause of stroke for those with anterior circulation infarcts but not for those with LACI or POCI. Subjects with PACI should be referred for early carotid imaging to identify those with severe disease who may be suitable for elective carotid surgery.