OBJECTIVES Previous cost-effectiveness studies have modeled Alzheimer's disease (AD) progression in terms of cognitive function alone, a single global severity measure, or progression to the need for “Full Time Care.” This study estimates AD progression in terms of correlated changes in cognition, behavior and function. These projections are then used to analyze the cost-effectiveness of donepezil versus standard care in the UK. METHODS Patient-level data from eight randomized placebo-controlled donepezil trials and a seven-year follow-up registry provided the basis for modeling longitudinal rates of change in cognition (MMSE), behavior (Neuropsychiatric Inventory), activities of daily living and Instrumental Activities of Daily Living. A discrete event simulation is used to project outcomes for two patient groups, identical except for treatment: donepezil 10mg per day versus untreated. Patient mix and costs were developed from UK-specific literature. Costs are reported in 2007 British pounds. The discount rate is 3.5%. RESULTS After ten years, patients with mild to moderately severe AD (26 ≥ MMSE ≥ 10) on donepezil are better off than those without treatment, with costs reduced from both health care system and societal perspectives, savings averaging ≤1421 and ≤4094 per patient respectively. Donepezil-treated patients experience 0.12 more QALYs per patient and their caregivers 0.01 QALYs compared to untreated patients and their carers. In sensitivity analyses, dominance holds over a wide range of inputs, including when treatment effects, the impact of disease severity on caregiver time and patient utility, and institutionalization rates are decreased by 25%; or when the time horizon is decreased to 5 years. In probabilistic sensitivity analyses, donepezil dominates in 53 to 77% of replications (depending on perspective), and results in cost/QALY estimates below ≤30,000 in 79 to 90% of replications. CONCLUSIONS These results suggest that donepezil is highly cost-effective in patients with mild to moderately severe AD in the UK.