BACKGROUND: In patients with moderately severe to severe Alzheimer’s disease, the N-methyl-D-aspartate (NMDA) antagonist memantine has been shown to improve outcomes and to be associated with reductions in resource utilisation and total healthcare costs relative to no pharmacological intervention after 28 weeks in phase III clinical and pharmacoeconomic studies. However, the longer term cost implications of treatment with memantine are not known. OBJECTIVE: To evaluate the effect of treatment with memantine in patients with moderately severe to severe Alzheimer’s disease on resource use and on cost and patient outcomes in Finland over a 5-year time horizon. METHODS: A Markov model was constructed to simulate a patient’s progression through a finite series of health states with a time horizon of 5 years. The states were defined in terms of physical dependency, place of residency (community or institution), and cognitive function. Each 6-month Markov cycle was repeated ten times. A 5% rate was used to discount costs. Inputs for the model were derived from epidemiological data collected during the Kuopio 75+ Study, a Finnish population-based health survey of dementia and functional capacity among in-dividuals aged >/=75 years. Costs were considered from a societal perspective. Probabilities used in the model, together with cost and resource use differences between treatment with memantine (Ebixa((R)), Namenda((R)), Axura((R))) and no pharmacological intervention, were derived from a randomised, double-blind, placebo-controlled clinical trial that included an economic assessment. This study enrolled 252 patients with moderately severe to severe Alzheimer’s disease. We took a conservative approach that assumed that the effectiveness of treatment with memantine was limited to 12 months’ duration. Monte Carlo simulations were performed to evaluate the effect of treatment with memantine on duration of independence and time to institutionalisation. Sensitivity analyses included memantine efficacy best- (5 years) and worst- (6 months) case scenarios, and an analysis in which 5% discounting was not applied. RESULTS: Memantine therapy was associated with approximately 4 extra months of independence, 1 additional month of residence in the community, and a cost reduction relative to placebo of approximately euro1700 per patient over 5 years, despite the limiting of persistence of efficacy to 12 months (year of costing, 2001). Monte Carlo simulations and sensitivity analyses supported the findings. CONCLUSION: According to the model, over 5 years the additional drug costs of treating patients with moderately severe to severe Alzheimer’s disease with memantine were amply offset by cost savings related chiefly to increased independence and delayed institutionalisation.