OBJECTIVES: To examine the direct medical costs of newly diagnosed patients with Alzheimer's disease (AD) using retrospective health care claims data. METHODS: This study examined individuals aged 65 years and over with pharmacy benefits who had at least one claim with an AD diagnosis and were enrolled in commercially-insured and nationally-dispersed Medicare Risk plans between January 1999 and November 2003. Each AD patient had an "index date" where the first AD claim was observed, a 12-month pre-index period, and a minimum 30-day follow-up. A control group consisted of individuals who had no AD or dementia over the study period and were randomly matched (2:1) to AD patients based on age, gender, and follow-up duration. The Charlson Comorbidity Index was used to examine the burden of comorbid medical conditions in the pre-index period. The primary measures of interest were annualized health care resource utilization and costs; a generalized linear model with a gamma distribution and log-link function was used to compare costs between the AD and control groups over the follow-up period. RESULTS: Both AD patients (n = 2475) and controls (n = 4950) were aged 82 years on average; 38% were male. AD patients had significantly more comorbid medical conditions than controls (mean Charlson score 1.6 vs. 1.2); the prevalence of diabetes, heart and vascular problems also was higher in the AD group. Inpatient costs contributed primarily to total annualized costs among AD patients, while outpatient costs dominated among controls. Average adjusted annualized costs for AD patients were more than five-fold higher compared to controls, driven primarily by inpatient costs ($21,150 vs. $4,053 for AD vs. control, respectively). CONCLUSIONS: AD patients have a significantly greater comoribidity burden than age- and gender-matched controls, and incur much higher levels of inpatient service use and overall health care costs.