Objectives To assess the value of non-invasive medical imaging for diagnosis of carotid artery stenosis and to study the relationship between carotid stenosis and brain infarction. Methods Thirty-one patients with a total of 62 carotid arteries were studied using Doppler ultrasound (DUS) and magnetic resonance angiography (MRA). Eleven of the 31 patients were studied using CT angiography (CTA). CT and MRI of the brain were also done in all patients. The imaging results in 5 patients were compared with those of digital subtraction angiography (DSA). Eight patients with severe stenosis received carotid endarterectomy. The comparisons between the imaging results and pathological data were conducted in 2 patients. Results Of the 62 carotid arteries, mild stenosis was seen in 11, moderate in 14, severe in 21, obstructed in 4 and normal in 12. In 25 patients with severe stenosis or occlusion of carotid arteries, there were a total of 35 focal or multifocal infarcts on the ipsilateral cerebral hemisphere, and 15 infarcts on the contrary side. Compared with the results of the operations, DUS correctly diagnosed 6 stenoses, while MRA identified 7 correctly and CTA 8. Agreement on location of stenosis as performed by endarterectomy, DUS, MRA and CTA occurred in all patients. Histologically, areas of calcification and fibrousness were related to high densities on CTA, strong echoes on DUS, and low signal intensities on MRA. Relatively large amounts of necrotic material and foam cells filled with lipolytic materials on the intimal surface of arteries were observed during pathologically, corresponding to low and iso-densities on CTA, low echoes on DUS, and inhomogeneous signal intensities on MRA. Conclusions A strong link exists between carotid stenosis and brain infarction. The combined use of DUS, MRA and CTA can improve diagnostic accuracy for the assessment of carotid artery stenosis, as well as assist in ascertaining the nature of the plaque.