High circulating levels of glucocorticoid hormones adversely affect cognition. Previous studies exploring the hypothalamic-pituitary-adrenal (HPA) axis and basal cortisol levels in the elderly reported that subjects with mid-range cortisol levels outperformed subjects with high cortisol levels on assessments of memory and attention. This study examines the efficacy of mifepristone, a glucocorticoid-antagonist, in decelerating the rate of cortisol-related cognitive decline in subjects with mile-to-moderate Alzheimer's disease (AD). Rate of cognitve decline is compared in AD subjects randomized to receive 200 mg of mifepristone daily for 6 mo or placebo. The Alzheimer's Disease Assessment Scale (ADAS) and the Folstein Mini Mental Status Exam (MMSE) will be the primary measures used to assess change in cognitve function over the 6 mo period, supplemented by a neuropsychological battery testing memory and language and reasoning skills. During each visit, subjects will have samples collected for determination of plasma adrenocorticotropin (ACTH), serum cortisol and salivary cortisol levels to assess HPA axis activity. The placebo arm of this study also investigate whether subjects with high baseline cortisol levels experience greater declines in cognitive impairment over time relative to subjects with Ad who have low baseline cortisol levels. Additionally, this study test the hypothesis that AD subjects with elevated cortisol at baseline will perform more poorly on neuropsychological exams that do subjects with low cortisol.