Objective and subjective assessments of postoperative improvement and impairment in cognition were prospectively compared in patients who underwent carotid endarterectomy (CEA). Each patient underwent subjective cognitive assessment by a neurosurgeon and the patient's next of kin, and neuropsychological testing consisting of five test scores within 7 days before surgery and between 1 and 2 months after surgery. Of 213 patients studied, 24 (11%), 166 (78%), and 23 (11%) patients were defined as having subjectively improved, unchanged, and impaired cognition, respectively, following surgery. In all neuropsychological tests, differences in test scores between the two tests (postoperative test score - preoperative test score) significantly differentiated patients with subjectively improved, unchanged, and impaired cognition after surgery. Receiver operating characteristic analysis showed that the cut-off point for the differences in neuropsychological test scores in detecting subjective improvement and impairment in cognition after surgery was identical to mean +2 standard deviations (SDs) and mean -2 SDs, respectively, of the control value obtained from normal subjects. Of 27 patients with differences in neuropsychological test scores more than the upper cut-off point and 26 patients with differences in neuropsychological test scores less than the lower cut-off point in one or more neuropsychological tests, 24 (89%) and 23 (88%) exhibited subjectively improved and impaired cognition, respectively, after surgery. The present study indicates that neuropsychological test scores reflect the subjective assessment of postoperative change in cognition, and can detect subjective improvement and impairment in cognition after CEA using the optimal cut-off points for the test scores.