Although both human epidemiologic and animal model studies have suggested that caffeine/coffee protects against Alzheimer's disease, direct human evidence for this premise has been lacking. In the present case-control study, two separate cohorts consisting of 124 total individuals (65-88 years old) were cognitively assessed and a blood sample taken for caffeine/biomarker analysis. Subjects were then monitored for cognitive status over the ensuing 2-4 year period to determine the extent to which initial plasma caffeine/biomarkers levels would be predictive of changes in cognitive status. Plasma caffeine levels at study onset were substantially lower (-51%) in mild cognitive impairment (MCI) subjects who later progressed to dementia (MCI→DEM) compared to levels in stable MCI subjects (MCI→MCI). Moreover, none of the MCI→DEM subjects had initial blood caffeine levels that were above a critical level of 1200 ng/ml, while half of stable MCI→MCI subjects had blood caffeine levels higher than that critical level. Thus, plasma caffeine levels greater than 1200 ng/ml (≈6 μM) in MCI subjects were associated with no conversion to dementia during the ensuing 2-4 year follow-up period. Among the 11 cytokines measured in plasma, three of them (GCSF, IL-10, and IL-6) were decreased in MCI→DEM subjects, but not in stable MCI→MCI subjects with high plasma caffeine levels. Coffee would appear to be the major or perhaps only source of caffeine for such stable MCI patients. This case-control study provides the first direct evidence that caffeine/coffee intake is associated with a reduced risk of dementia or delayed onset, particularly for those who already have MCI.