We report 91 patients (mean age 70 years) operated upon, prospectively for a total of 100 carotid revascularizations (nine bilateral). Eighty-five of these patients had pre-, intra-, and postoperative transcranial Doppler investigations. Preoperatively, these 85 patients (92 procedures) were classified into two groups based on the results of their Doppler examinations: Group A (65 patients, 72 procedures), those who did not require an intraoperative indwelling shunt and Group B (20 patients, 20 procedures), those who did. The shunt was inserted only when the mean stump (back) pressure was less than 50 mmHg after cross-clamping. Group A all had satisfactory collaterality with a functional anterior and one or two posterior communicating arteries. Group B had no communicating arteries (anterior or posterior) identified by transcranial Doppler. In 17 of 20 patients in this group, the stump pressure was less than 50 mmHg and a shunt was placed. The overall prediction based on Doppler examination of whether or not patients would need a shunt during operation for the two groups A and B (i.e., 92 procedures) was correct in 95.6% (88/92) of cases. Moreover, six hemodynamically significant stenoses (four in the cavernous portion, two in the middle cerebral artery) were disclosed. Sensitivity and specificity of transcranial Doppler as correlated with arteriographic findings were 70 and 90%. Preoperative transcranial Doppler can measure the velocities of the principal cerebral arteries and the collateral capacity of the circle of Willis, and can forecast tolerance to carotid cross-clamping. Intraoperatively, the velocity of flow in the middle carotid artery was correlated with stump pressure, which allowed for surveillance of the shunt.