The authors estimated the effects of each of the three commonly used drugs for Alzheimer disease (donepezil, galantamine, and rivastigmine) in terms of predefined clinical outcomes and trial completion rates, by dosing level, and described differences among them. Using both electronic and manual search strategies (January 1992 to July 2002), a metaanalysis examined the effect of the drugs on clinical outcomes and completion rates. Regression analyses compared the effect of dose on clinical outcomes and completion rates, using 10 donepezil, 6 galantamine, and 5 rivastigmine articles. All three drugs showed beneficial effects on cognitive tests, as compared with placebo. For donepezil and rivastigmine, larger doses were associated with larger effect. This was not the case with galantamine. The odds of clinical global improvement demonstrated superiority over placebo for each drug, with no dose effects noted. Dropout rates were greater with galantamine and rivastigmine. There was little difference in dropout rate for each drug at each dose-level, except with high-dose donepezil. This was accounted for by the high dropout rate in two 52-week studies using larger doses. In summary, all three drugs had similar cognitive efficacy, with donepezil and rivastigmine showing a dose effect across the dosing levels studied. However, both galantamine and rivastigmine were associated with a greater risk of trial dropout than placebo, especially at higher dosing levels.