Background Alzheimer's disease is the most common cause of dementia in older people. One of the aims of therapy is to inhibit the breakdown of a chemical neurotransmitter, acetylcholine, by blocking the relevant enzyme. This can be done by a group of chemicals known as cholinesterase inhibitors. Objectives The objective of this review is to assess whether donepezil improves the well-being of patients with dementia due to Alzheimer's disease. Search methods The Cochrane Dementia and Cognitive Improvement Group's Specialized Register was searched using the terms 'donepezil', 'E2020' and 'Aricept' on 5 April 2006. This Register contains up-to-date records of all major health care databases and many ongoing trial databases. Members of the Donepezil Study Group and Eisai Inc were contacted. Selection criteria All unconfounded, double-blind, randomized controlled trials in which treatment with donepezil was compared with placebo for patients with mild, moderate or severe dementia due to Alzheimer's disease. Data collection and analysis Data were extracted by one reviewer (JSB), pooled where appropriate and possible, and the pooled treatment effects, or the risks and benefits of treatment estimated. Main results 24 trials are included, involving 5796 participants, of which 15 reported results in sufficient detail for the meta-analyses. Most trials were of 6 months or less duration in selected patients. Patients in 20 trials had mild to moderate disease, in two trials moderate to severe, and in two severe disease. Available outcome data cover domains including cognitive function, activities of daily living, behaviour, global clinical state, adverse events and health care resource costs. For cognition there is a statistically significant improvement for both 5 and 10 mg/day of donepezil at 24 weeks compared with placebo on the ADAS-Cog scale (-2.01 points MD, 95%CI -2.69 to -1.34, P < 0.00001); -2.80 points, MD 95% CI -3.74 to -2.10, P < 0.00001) and for 10 mg/day donepezil compared with placebo at 24 weeks (5.55 SIB points, 95% CI 3.60 to 7.49, p<0.00001) and 52 weeks (1.84 MMSE points, 95% CI, 0.53 to 3.15, P =0.006). The results show some improvement in global clinical state (assessed by a clinician) in people treated with 5 and 10 mg/day of donepezil compared with placebo at 24 weeks for the number of patients showing improvement (OR 2.38, 95% CI 1.78 to 3.19, P = < 0.00001, OR 1.82, 95% CI 1.42 to 2.35, P < 0.00001). Benefits of treatment were also seen on measures of activities of daily living and behaviour, but not on the quality of life score. There were significantly more withdrawals before the end of treatment from the 10 mg/day (289/1125 24% 10 mg/day vs 219/1079 20% placebo, OR 1.35, 95% CI 1.11 to 1.65, P = 0.003) but not the 5 mg/day, (100/561 18% 5 mg/day vs 109/568 19% placebo, OR 0.91, 95% CI 0.68 to 1.24, P = 0.56) donepezil group compared with placebo which may have resulted in some overestimation of beneficial changes at 10 mg/day. Benefits on the 10 mg/day dose were marginally larger than on the 5 mg/day dose. The results were similar for all severities of disease. Two studies presented results for health resource use, and the associated costs. There were no significant differences between treatment and placebo for any item, the cost of any item, and for the total costs, and total costs including the informal carer costs. Many adverse events were recorded, with more incidents of nausea, vomiting, diarrhoea, muscle cramps, dizziness, fatigue and anorexia (significant risk associated with treatment) in the 10 mg/day group, compared with placebo. There were more incidents of anorexia, diarrhoea, and muscle cramps in the 5 mg/day group compared with placebo, but not of dizziness, fatigue, nausea or vomiting. Very few patients left a trial as a direct result of the intervention. Authors' conclusions People with mild, moderate or severe dementia due to Alzheimer's disease treated for periods of 12, 24 or 52 weeks with donepezil experienced benefits in cognitive function, activities of daily living and behaviour. Study clinicians rated global clinical state more positively in treated patients, and measured less decline in measures of global disease severity. There is some evidence that use of donepezil is neither more nor less expensive compared with placebo when assessing total health care resource costs. Benefits on the 10 mg/day dose were marginally larger than on the 5 mg/day dose. Taking into consideration the better tolerability of the 5 mg/day donepezil compared with the 10 mg/day dose, together with the lower cost, the lower dose may be the better option. The debate on whether donepezil is effective continues despite the evidence of efficacy from the clinical studies because the treatment effects are small and are not always apparent in practice.