Preliminary results of the North American Symptomatic Carotid Endarterectomy Trial (NASCET) have shown that endarterectomy is highly effective in preventing stroke in patients with 70% to 99% stenosis of the carotid bifurcation. These results are based on specific linear measurement criteria comparing the residual diameter of the lumen with that of the normal carotid beyond the bulb. These criteria may differ from subjective assessments of the severity of stenosis. Angiograms from 125 NASCET patients were analysed independently by three groups of reviewers. The first group consisted of three neuroradiologists, the second of two neurologists and one neurosurgeon and the third of three radiologists with neuroradiologic experience. The degree of stenosis was estimated, without measurement, as mild (1% to 29%), moderate (30% to 69%) or severe (70% to 99%), or the vessel was said to be occluded. These assessments were compared with the degree of stenosis as determined by the NASCET neuroradiologist on the basis of measurement. The subjective assessment of stenosis produced false-positive rates (for assessing stenosis of less than 70% as severe) ranging from 5% among the neuroradiologists to 16% among the radiologists. The false-negative rates (for assessing stenosis of 70% or more as less than severe) ranged from 12% to 22%. Therefore, a subjective rating of carotid stenosis may lead to erroneous estimation of the severity of disease. This problem has therapeutic implications, because carotid endarterectomy has been shown to benefit only patients with stenosis of greater than 70%, as assessed according to strict measurement guidelines.