The cognitive subscale of the Alzheimer's Disease Assessment Scale (ADAS-Cog) is used to monitor disease progression and treatment efficacy in clinical trials of Alzheimer's disease (AD). Using data from a 12-week drug trial, we retrospectively studied the effect of education on ADAS-Cog performance in a group of 444 patients with AD. The effect of education was statistically significant on baseline ADAS-Cog total scores. This effect remained statistically significant after controlling for age, gender, and dementia severity. Education effects were also statistically significant at week 12 for ADAS-Cog total and 10 of 11 subitem scores in 138 AD patients in the placebo arm of the trial. Post hoc analysis showed that non-high school graduates performed worse than those with greater educational levels across a broad range of cognitive domains. Our results, in conjunction with reports linking lower educational level with a higher risk for AD, suggest that educational level of patients be given consideration in the design and interpretation of cognitive tests in AD drug trials.