The association between cholesterol levels, cognitive impairment and dementia is controversial. This study aimed to investigate 1) differences in serum total cholesterol levels between mild and moderate to severe cognitive impairment, and between dementia syndromes; 2) the association between cholesterol level and Alzheimer's disease (AD); and 3) the association between cholesterol level and progression of AD. Cholesterol levels were measured in two groups: 1) 341 participants in a community study in Gwangju, aged 65 or over, scored below 25 on the Korean version of Mini-Mental State Examination (MMSE-K); and 2) 79 patients with AD attending a local hospital. In the community sample, this association was investigated between cholesterol level and both cognitive function and dementia. For the hospital sample, associations were investigated between cholesterol level and decline in cognitive function (MMSE-K) over one year. Lower serum cholesterol level was associated with worse cognitive function in the community sample. This association was specific to AD rather than other subtypes. No cross-sectional association was found between cholesterol levels and cognitive function in AD groups from either sample. No prospective asso- ciations were found between cholesterol level and decline of cognitive function in hospital attenders with AD. Adjustment for age, gender, education, past occupation, disablement, duration of dementia, and the presence of apolipoprotein E e4 made little difference to the associations. These findings suggest that lower serum cholesterol level may be a state marker of AD but does not appear to influence its rate of progression.