It is biologically plausible that hormone replacement therapy (HRT) would be protective against cognitive decline and Alzheimer's disease (AD). We review observational and randomized trials to determine whether HRT might protect against cognitive decline in cognitively unimpaired and demented women. We also address issues of clinical relevance, including duration and type of treatment and patient characteristics, including type of menopause (surgical versus natural), age, education and menopausal symptoms. Differences in participant characteristics and testing methods limit the ability to draw conclusions across randomized studies of HRT in non-demented women. The available evidence suggests no detrimental effect of HRT on cognitive function and inconsistent benefits on verbal memory and reasoning, frontal functions and speeded attention. Meta-analyses of observational trials suggest that HRT protects against the development of AD, but randomized trials indicate no long-lasting benefit in patients with AD. Evidence is insufficient to recommend HRT to maintain cognitive function.