A radioisotopic method of quantifying mean cerebral transit time was used to assess the immediate effects of carotid endarterectomy on cerebrovascular reserve (CVR) in 69 patients. In addition, serial postoperative data were acquired on CVR, clinical status and non-invasive imaging of the internal carotid arteries in 56 patients over a period of 6–48 (median 24) months. Twenty-one patients (30 per cent) had preoperative evidence of impaired CVR in the symptomatic hemisphere. Within 4 days of surgery, however, reserve had returned to normal in 17 of the 21 patients. During follow-up, four of the 56 patients developed recurrent stenosis (< 50 per cent) or occlusion of the artery operated on but only two of these had impairment of CVR and none was symptomatic. Three patients suffered recurrent transient ischaemic attacks (TIAs) but none had recurrent internal carotid artery disease or impaired CVR. One patient suffered a TIA in the territory of the non-operated artery during follow-up in association with disease progression and CVR impairment. However, the TIA preceded recognition of either of these changes. Twelve other patients had (or developed) stenosis ( > 50 per cent) in the non-operated artery during follow-up but none was symptomatic or developed impairment of CVR. Although assessment of CVR provided useful information on the frequency of haemodynamic compromise before carotid endarterectomy and on the natural history of disease progression, neither serial assessment of reserve nor non-invasive imaging of the carotid bifurcation influenced clinical practice during follow-up.