Deep hypothermia with circulatory arrest is the usual method of cerebral protection during replacement of the aortic arch. It has the enormous advantage of allowing the surgical repair to be carried out in a complete bloodless field with no aortic cross-clamping. However, this method only gives the surgeon a limited period of time to carry out the aortic repair. It also requires that cardiopulmonary bypass be prolonged to cool and rewarm the patient which may be the cause of various complications. It has been proposed to improve the efficiency and the results of deep hypothermia, by associating it with retrograde cerebral perfusion of the brain with oxygenated blood through the superior vena cava. This technique improves the tolerance of the brain to cold ischemia and increases the time of repair allowed to the surgeon. Antegrade selective cerebral perfusion has also been in use for more than three decades. When the perfusion is derived from the main arterial line and performed at moderate hypothermia, the aorta must be cross-clamped to perform the repair. In addition, there is some uncertainty as to what constitutes adequate perfusion flow at normal or moderate hypothermic conditions. To reconcile the advantages of both approaches while avoiding their major drawbacks, in 1986 we proposed an original method of selective antegrade brain perfusion. The principle is to perfuse selectively the brain with cold blood (10 to 12 degrees C) while maintaining the central temperature in moderate hypothermia (25-28 degrees C). During the time of the distal anastomosis the cardiopulmonary bypass is stopped, maintaining only the cerebral perfusion at a flow rate of about 400 to 500 mL/mn and a pressure of about 70 mmHg. As soon as the distal anastomosis is completed the main perfusion is resumed. Two hundred and six patients with a mean age of 57 years (22 to 83) were operated on with this technique between October 1984 and March 2001. One hundred forty three patients underwent an elective procedure and 63 patients were operated on in emergency, mainly for acute type A dissection (54 of 63). The hospital mortality was 17% (34 patients). Death was directly related to neurological injury in 9 patients (4.4%). All others patients awoke within 6 to 8 hours and were conscious at 24 hours postoperatively. Thirteen nonfatal neurological complications were observed. The type of lesion, gender, age, duration of CPB, cerebral perfusion, and circulatory arrest had no influence on the neurological outcome of the patients. In our experience, antegrade selective perfusion of the brain with cold blood and moderate hypothermic central temperature constitutes the method of choice for cerebral protection during surgery of the aortic arch as it requires no prolonged CPB and does not limit the time available to perform the aortic repair.