OBJECTIVES: To review the cost-effectiveness of cholinesterase inhibitors (ChEIs; donepezil, galantamine and rivastigmine) and memantine for treating patients with AD. This review aims to inform policy makers in Singapore on the published cost-effectiveness studies of these drug therapies. METHODS:  Electronic databases (EMBASE, MEDLINE, NHS EED) were searched from inception to February 2010 for published economic evaluation studies. Additional references were identified through searching bibliographies of included studies and related publications. Review papers of economic evaluations either on individual drugs or more of these drugs were considered. An outline assessment of economic evaluations was undertaken using Drummond’s checklist.  RESULTS:  A systematic search of the literature identified 32 economic evaluations (2 of all three ChEIs, 11 of donepezil, 5 of rivastigmine, 7 of galantamine and 7 of memantine). All studies used clinical trial data from single drug comparisons with placebo. Different approaches have been used to model disease progression for evaluating cost-effectiveness of each drug, making comparison between drugs more difficult. Caro et al (2003) and Green et al (2005) considered the three ChEIs using a common analytical framework. Most of the studies reported that drugs provided health benefits and cost savings over time, suggesting that their use was cost-effective. However, Green et al (2005) concluded that treatment with ChEIs may not be cost effective from the UK NHS’s perspective. CONCLUSIONS:  The majority of economic evaluations of ChEIs and memantine reported that these treatments were the dominant strategies compared with no treatment. However, conclusions from the reviews of economic analyses were mixed, and the authors commented that the findings from economic analyses of these four drugs should be interpreted with caution due to the assumptions and methodology used. Comparison between studies is difficult due to difference in study types, geographic locations, perspectives and different assumptions within the models.