OBJECTIVES: Placebo-controlled clinical trials have demonstrated the efficacy of memantine versus placebo in moderately severe and severe Alzheimer's disease. A modelling approach has been adopted to estimate costs and outcomes of memantine treatment in clinical practice in Spain. METHODS: A Markov model simulated moderately severe and severe patients' progression through levels of combinations of severity, autonomy, and setting over five years. Model inputs include clinical trial results and measurement of patient and caregiver resource utilisation from a societal perspective using literature and expert data. The main outcome measures were time spent in autonomy (patient's ability to accomplish activities using the ADCS-ADL scale) and time to institutionalisation. This cost-effectiveness study compared the memantine treatment strategy to that of no pharmacotherapy and to donepezil for patients with moderately severe AD followed by no pharmacotherapy when patients reach the severe AD level. RESULTS: The time spent in autonomy for patients treated with memantine is 14% longer in duration than for patients treated with donepezil and 25% longer than for patients without pharmacotherapy. The delay time to institutionalisation was 8% longer with memantine compared with donepezil and 12% longer with memantine compared with no treatment. Over a 5-year period, patients treated with memantine showed a decrease of €268 and €511 in total Health care costs, compared with donepezil and no treatment, respectively. Robustness of the results was validated by a comprehensive sensitivity analysis. CONCLUSION: Memantine is more effective in increasing time spent in autonomy and in reducing Health care costs compared with donepezil or no pharmacotherapy.