We feel that the paper by Bengt Winblad and colleagues1 on the use of donepezil for patients with severe Alzheimer's disease omits two essential points which should be considered when interpreting this study. The first point is that there is now increasing evidence that donepezil has more than one mechanism of action. As well as being an acetylcholinesterase inhibitor, there is mounting support for the idea that donepezil, and also galantamine, have additional neuroprotective activity. Clinical studies2 and 3 have shown that use of these drugs at an early stage results in delayed progression of the disease, and increased longevity, implying a disease-modifying effect rather than the simple suppression of symptoms that might be expected from acetylcholinesterase inhibition. Basic research4 suggests that this neuroprotective effect is due to the ability of these drugs to inhibit the aggregation and toxicity of the β-amyloid peptide that accumulates within senile plaques. The second point is related to the first in that donepezil is most effective in delaying disease progression in patients with the apolipoprotein E 4 genotype,5 who have very high levels of senile plaque formation. We welcome this study and strongly agree with Winblad and colleagues' view that donepezil is of benefit to these patients. In our experience, even small improvements make a real difference to the quality of life of patients and carers. We declare that we have no conflict of interest.