Since October 1984 to December 1992, 47 aortic arch replacements were performed under selective cerebral perfusion (SCP). Patient age ranged from 19 to 81 years, with an average of 60.2 years. Thirty one patients were male and 16 female. Twenty patients were diagnosed with dissecting aneurysms, 27 true aneurysms. The operative mortality was 14 out of 47 (29.8%) over all. The mortality of dissecting aneurysm related acuity that is 5 of 9 (55.6%) for patients in acute phase, against none in chronic. The operative mortality of true aneurysms was 6 out of 11 (54.5%) for patients who had emergency operation, against 3 out of 16 (18.8%) scheduled, extensive surgery. Cerebral disorders developed on 8 patients out of 44 (18.2%, except 3 operative deaths). The mortality rate of cerebral disorder was 6 out of 8 (75.0%). Six patients with true aneurysm had cerebral infarctions after the surgery. Two patients had already cerebral disorders in advance the surgery. The cause of cerebral infarction was resumed by atherosclerotic materials. Our operative results indicate that cerebral disorder subsequently affects the mortality of aortic arch replacement. For the prevention of cerebral infarction, we modified the method of SCP. We start the SCP at the initiation of total cardiopulmonary bypass, routinely associated to an aortic clamp and cardiac arrest under myocardial protective procedure that is the use of perfusion cooling (30 degrees C, monitored at the esophagus), topical cooling and cold Saint Thomas solution infused in the coronary arteries. When aortic arch replacement is done, we need to make sure a good restarting of the heart and a good recovery of the systemic circulation.(ABSTRACT TRUNCATED AT 250 WORDS)