The relation between self-reported cognitive dysfunction and neuropsychological performance over 24 weeks was assessed in a sample of 53 multiple sclerosis patients. Subjects were assessed at Weeks Zero and 24 as part of a clinical trial to enhance cognition. At baseline, subjects had at least mild cognitive impairment on the Rey Auditory Verbal Learning Test and an absence of depression. Neuropsychological performance was assessed with a modification of the well standardized Brief Repeatable Battery. The 5-item Perceived Deficits Questionnaire and a 2-item memory and attention/concentration questionnaire assessed self-perceived cognitive impairment. Self-assessed cognition did not correlate with neuropsychological performance at either baseline or 24 weeks. However, changes in the self-assessment measures did correlate with changes in neuropsychological performance. Patients accurately perceived some changes in their level of cognitive dysfunction, though they were insensitive to the degree of their current dysfunction. Possible explanations of this pattern of results are discussed. (JINS, 2005, 11, 614–619.)