We investigated the relationship between animal product consumption and evidence of dementia in two cohort substudies. The first enrolled 272 California residents matched for age, sex, and zip code (1 vegan, 1 lacto-ovo-vegetarian, and 2 ''heavy'' meat eaters in each of 68 quartets). This design ensured a wide range of dietary exposure. The second included 2,984 unmatched subjects who resided within the Loma Linda, California area. All subjects were enrolled in the Adventist Health Study. The matched subjects who ate meat (including poultry and fish) were more than twice as likely to become demented as their vegetarian counterparts (relative risk 2.18, p = 0.065) and the discrepancy was further widened (relative risk 2.99, p = 0.048) when past meat consumption was taken into account. There was no significant difference in the incidence of dementia in the vegetarian versus meat-eating unmatched subjects. There was no obvious explanation for the difference between the two substudies, although the power of the unmatched substudy to detect an effect of ''heavy'' meat consumption was unexpectedly limited. There was a trend towards delayed onset of dementia in vegetarians in both substudies.