To show relationship between degree of carotid arterial stenosis and cerebral blood flow reactivity (RES%) to induced hypercapnia, fluorine-18-fluoromethane and positron emission tomography (PET) was used to study 18 patients with carotid distribution transient ischaemic attacks (TIA), all free of stroke, who had angiographic-proven unilateral arterial disease. Non-involved carotid arteries were either normal or had non-stenotic plaque. Either normal arteries or nonstenotic ulcerations in the symptomatic carotid arteries were present in five of 18 (28%), ipsilateral carotid stenosis from 50-99% was present in eight of 18 (44%), and ipsilateral internal carotid occlusion was present in five of 18 (28%) patients. In comparison with 14 normal controls, all patients with symptomatic middle cerebral artery (MCA) flow territories had significantly lower mean (SEM) RES% [5.0' (0.2) vs 4.0 (0.9), p less than 0.04]. Symptomatic anterior borderzone (ABZ) RES% was also significantly lower [4.6 (0.4) vs 3.3 (0.9), p less than 0.04], than controls. In patient subgroup comparisons, the 50-99% stenosis subgroup clearly had the lowest MCA RES% [3.4 (0.2)] as well as the lowest ABZ RES% [2.8 (0.4)] on their symptomatic sides. Age, expired pCO2, mean arterial blood pressure, serum glucose, serum haematocrit and number, type and estimated duration of TIAs were not significantly different between subgroups. Linear regression showed a significant relationship between RES% and both measured percentage-stenosis (p = 0.04) and residual luminal diameter (p = 0.05) in symptomatic MCA territories. This approached significance in symptomatic ABZ regions. This preliminary data set suggests that unilateral carotid stenosis can and does result in impaired CO2 reactivity following hypercapnia. The relative normality of CO2 reactivity in those with carotid occlusion is discussed.