Low testosterone (T) levels may predispose to Alzheimer disease (AD), but it is unclear whether this is a comorbid effect due to cachexia, subclinical hyperthyroidism or other co-morbidity. The biological plausibility for potential protective effects of T on brain functions is substantial. In addition, higher levels of gonadotropins found in older cases with AD suggest that low levels of T are not due to brain degeneration and that the hypothalamic-pituitary-gonadal (HPG) axis is still intact. Men genetically at risk for AD were also already found to have lower levels of T. However, despite having lower levels of T, women do not show accelerated cognitive decline with age when compared to men. In addition, castration has not necessarily shown a decline in cognitive functions; some studies even found improvement of memory recall. Age may be an important factor when assessing optimal levels of T and several studies suggest that free or bioavailable T may be a better marker than total T levels when investigating associations of androgen activity with cognitive function. Small-scale T intervention trials in elderly men with and without dementia suggest that some cognitive deficits may be reversed, at least in part, by short term T supplementation. Age and prior hypogonadism may play an important role in therapy success and these factors should be investigated in more detail in future large scale randomized controlled studies. For elderly women, T treatment does not seem to have additional benefits over estrogen treatment for postmenopausal complaints and cognitive decline and may increase cardiovascular disease.