OBJECTIVES Assess the cost-effectiveness in a UK setting of providing memantine treatment to moderate to severe Alzheimer's disease (AD) in patients receiving stable donepezil treatment compared with not providing memantine. METHODS Data from a U.S. multicenter randomised clinical trial that compared memantine versus placebo in moderate to severe AD patients on stable doses of donepezil was used to evaluate the cost effectiveness of providing memantine to donepezil. Using methodology developed by Kurz et al., patients were classified at each visit as dependent or non-dependent according to their ADCS-ADL19 scores. Mean weekly costs were estimated from the National Health Service and Personal Social Services perspectives for patients with MMSE £ 14 that participated in a UK epidemiological study - the LASER-AD study. QALYs per dependency were estimated from a Danish Study. Per week mean acquisition cost and effectiveness of treatment were £ 975; 0.6511 and £ 288; 0.3207 for dependent and non-dependent patients. Total drug costs for the 24-week period were £ 492 for donepezil and £ 937 for memantine plus donepezil. Costs, QALYs and time of non-dependence were associated with each dependency level and added to obtain total outcomes over the 24-week study period. RESULTS Over the evaluation period, memantine added to patients stabilised on donepezil was associated with an additional 0.0112 QALYs, an additional 1.77 weeks of independence and a £ 771 cost reduction compared with donepezil alone. The cost reduction is not statistically significant but suggests that clinical advantages offset some of the cost of adding memantine. CONCLUSION This analysis suggests that memantine treatment provided to patients receiving stable donepezil treatment is cost-effective compared with not adding memantine. As costs and QALYs were assessed retrospectively, further prospective studies are required to support this finding.