OBJECTIVE The efficacy and safety of memantine in patients with moderate-to-severe Alzheimer's disease (AD) receiving stable doses of donepezil were recently demonstrated in a phase III trial. The cost-effectiveness of such therapy is unknown. METHODS A microsimulation model was developed to depict AD progression over time and associated clinical and economic outcomes. AD progression was measured in terms of decline in cognitive function, as assessed by the Severe Impairment Battery (SIB). At model entry, patients were assumed to have moderate-to-severe AD, to be on stable doses of donepezil, and to begin combination therapy with memantine, or continue to receive donepezil alone; duration of therapy was assumed to be 1 year. Drug efficacy was based on data from a phase III trial. Key assumptions of the model included: (1) efficacy of study drugs would extend to 1 year; (2) measures of cognitive function could be mapped to one another, as well as to global measures of disease severity; and (3) following therapy discontinuation, cognitive function would revert immediately to natural history levels. Cost-effectiveness was assessed in terms of cost (2005 US$) per quality-adjusted life-year (QALY) gained over a lifetime (3% discount rate). RESULTS SIB scores were estimated to improve by 3.3 over 1 year from therapy with memantine plus donepezil (vs. donepezil alone). While pharmacotherapy costs were estimated to increase by $1250 during the year of memantine treatment, costs of formal and informal services were estimated to decrease by $1240 over this period and by $1493 (discounted present value) over a lifetime. Findings were sensitive to the assumed SIB score at therapy initiation; cost-effectiveness was better for patients with higher initial SIB scores (i.e., less severe disease). CONCLUSIONS In patients with moderate-to-severe AD already receiving donepezil, treatment with memantine results in improved clinical outcomes and reduced total costs of care.