Explanations for the association between educational attainment and the risk of dementia fall into three main categories. It may arise as an artefact of study methods; education may predict broader socioeconomic circumstances and exposures, or education may reflect brain reserve or cognitive capacity that protect against dementia. Data from the Canadian Study of Health and Aging (N=6646, giving 44,676 person-years of follow-up) are analyzed to test a series of hypotheses reflecting these explanations. Years of education showed a strong association with the risk of dementia (relative risk [RR] 2.1 for those with less than 6 years of education compared to those with 13 or more years; RR=2.9 among survivors). Possible artefactual factors were detected, but were insufficient to invalidate the association. Adjustments for a range of other socioeconomic indicators, health problems and lifestyle factors reduced, but did not remove, the association. Adjustments for intelligence and for an indicator of lifetime mental activity also appeared to account for some but not all of the association. The conclusion is that there appears to be a real association between educational attainment and the risk of dementia 50 to 60 years later; this influence appears to run through a number of different, and sometimes complementary, pathways.