OBJECTIVES: In light of recent clinical evidence, the indication of the NMDA antagonist memantine has been extended to “moderate to severe Alzheimer's disease (AD)”. No pharmaco-economic evaluation has been performed in this indication to date. This study provides an estimate of the cost-effectiveness of memantine compared with standard care (no pharmacotherapy) in moderate to severe AD adapted to a Canadian setting and including all available evidence. No other pharmacological treatment was included in the evaluation as memantine is currently the only drug approved in this indication. METHODS: The progression of AD in terms of cognitive severity, functional disability and mortality was simulated over two-years using a state-transition (Markov) model. Outcomes of the model were Quality-Adjusted Life-Years (QALY) and costs from a societal perspective. The main cost and epidemiological input parameters of the model were computed using data from the Canadian Study on Health and Aging (CSHA). All relevant published and unpublished clinical trials of memantine versus placebo in moderate to severe AD were used to compute the transition probabilities between health states. A priori distributions were associated to all relevant parameters in order to enable stochastic analyses. RESULTS: Compared with standard care, the memantine strategy produced 0.03 additional QALYs, with no additional overall cost. Probabilistic sensitivity analyses give 83.3% chance that memantine treatment is cost neutral, 89.5% chance of being cost-effective if the decision-maker is willing to pay $20,000 for a quality-adjusted life year and 96.2% chance for a willingness-to-pay of $100,000 per QALY. Robustness of the results was confirmed through one-way and scenario-based sensitivity analyses. CONCLUSION: Our evaluation found memantine dominant over standard care. Results were comparable with those published for acetylcholinesterase inhibitors indicated for treatment of earlier stages of AD.