The results of in vitro and animal studies provide a strong rationale for the use of hormone replacement therapy (HRT) to prevent dementia and Alzheimer's disease (AD). In humans, the results of 16 observational studies are consistent with the hypothesis that estrogen use reduces the risk of AD by 10 to 60%. However, women who are prescribed HRT are less likely to have hypertension, diabetes and history of stroke than nonusers. As all of these factors have been associated with increased risk of dementia (including AD), this “prescription bias” may have a significant impact on the results of observational studies. Randomized trials are designed with the aim of avoiding many of the potential biases and confounding (measured or unmeasured) of observational studies. The results of the Women's Health Initiative Memory Study (WHIMS) indicate that HRT (estrogen plus progestin or estrogen alone) increases the risk of dementia (hazard ratio, HR=1.8, 95% CI=1.2–2.6). Taking into account the results of the WHIMS and the adverse health events associated with the use of estrogen plus progestin or estrogen alone, we conclude that HRT cannot be recommended as a safe and effective strategy to prevent dementia.