Dementia incidence increases exponentially with age even in people aged 90 years and above. Because therapeutic regimens are limited, modification of lifestyle behaviors may offer the best means for disease control. To test the hypotheses that lifestyle factors are related to lower risk of dementia in the oldest-old, we analyzed data from The 90+ Study, a population-based longitudinal cohort study initiated in 2003. This analysis included 587 participants (mean age=93 y) seen in-person and determined not to have dementia at enrollment. Information on lifestyle factors (smoking, alcohol, caffeine, vitamin supplements, exercise, and other activities) was obtained at enrollment and was available from data collected 20 years previously. After an average follow-up of 36 months, 268 participants were identified with incident dementia. No variable measured 20 years previously was associated with risk. Engagement in specific social/mental activities and intakes of antioxidant vitamin supplements and caffeine at time of enrollment were, associated with significantly reduced risks. When these variables were analyzed together, the HRs changed little and remained significant for reading (0.54, P=0.01) and going to church/synagogue (HR=0.66, P<0.05) but not for caffeine (HR=0.61, P=0.15) and vitamin C (HR=0.68, P=0.07). While lifestyle behaviors around age 70 did not modify risk of late-life dementia, participation in activities and caffeine and supplemental vitamin intake around age 90 warrant further investigation.