OBJECTIVE: To examine the acute adverse outcomes and direct health care costs among patients with Alzheimer's disease (AD) using a retrospective, administrative claim database. METHODS: We identified an over-age-65 population with pharmacy and medical benefits enrolled in a large, US, geographically diverse, commercial managed care plan between May 2001 and December 2002. AD patients had at least one claim with an AD diagnosis or one filled prescription for medication used exclusively for AD treatment. This claim identified the index date. A control cohort consisted of non-AD patients with no dementia diagnosis over the pre- and post-index periods randomly matched (3:1) to the AD patients by age, gender, plan location, and length of enrollment. The first claim in the period identified the index date. All patients included in the study had a 12-month pre-index period, and a minimum of 30-days follow-up. We compared the prevalence of acute adverse outcomes and comorbidities between the AD and control cohorts. Additionally, we used a two-part model (one equation estimating the probability of any costs, and a generalized linear model with a gamma distribution and log-link function estimating the level of costs) to examine differences in adjusted annualized total health care costs between the AD patients and the controls. RESULTS: Both the AD patients (N = 4,550) and the controls (N = 13,650) had a mean age of 79 years. Approximately 70% of AD patients were identified based on an AD prescription. AD patients had a higher risk of fracture, accidental fall, and urinary tract infection than the controls. Annual adjusted total health care costs per patient were approximately $1418 greater for the AD cohort. CONCLUSIONS: AD patients had significantly greater risk of acute adverse outcomes and more health care resource utilization than age- and gender-matched controls in a large managed care plan.