Objective: Retrograde cerebral perfusion (RCP) is commonly used in thoracic aortic surgery, ostensibly to provide metabolic support, maintain cerebral hypothermia and/or wash out particulate emboli. We tested the hypothesis that RCP would affect neuropsychological outcome in a clinical cohort. Methods: Ninety-four patients undergoing elective thoracic aortic repairs requiring deep hypothermic circulatory arrest consented to participate in this study. These patients underwent preoperative neuropsychological evaluation and comprise the reference group. Fifty-six of these patients also underwent neuropsychological evaluation several weeks postoperatively, 12 of whom (21%) had RCP. The neuropsychological domains tested were attention, processing speed, memory, executive function, and ®ne motor function. A global assessment of impairment, negative neuropsychological outcome (NNO), was de®ned as a postoperative decrease in function in two or more neuropsychological domains for patients with at least three domains tested both pre-and postoperatively (n 48). The relationship of three potential predictors (RCP, cerebral ischemia time and patient age) to negative outcomes was analyzed using Wilcoxon two-sample tests, x 2 tests, Mantel±Haenszel tests and multiple logistic regression. P , 0:05 was considered signi®cant. Results: Memory dysfunction and NNO had strong associations with RCP. This effect remained signi®cant when controlling separately for age and cerebral ischemia time. Conclusions: The effects of RCP are dif®cult to distinguish from those of age and prolonged cerebral ischemia time, because complex thoracic aortic repairs are associated with advanced age, prolonged cerebral ischemia and use of RCP. Despite this limitation, these preliminary data indicated that RCP had no bene®cial effect (and most likely a negative effect) upon cognitive outcome. q