Strategies to prevent or delay nursing home admission in individuals with cognitive impairment are urgently needed. We hypothesized that physical inactivity, not consuming alcohol (as opposed to moderate alcohol use), and having a history of smoking predict nursing home admission among individuals with normal cognitive function, but these behavioral factors would have attenuated associations with nursing home admission among individuals with impaired cognition. We performed a prospective cohort study among 7631 Health and Retirement Study participants aged 65+ at baseline. Baseline dementia risk (high versus low, based on brief psychometric assessments and proxy reports) and modifiable risk factors (physical inactivity, ever smoking, and not consuming alcohol) were used to predict nursing home admission in pooled logistic regression models. We evaluated whether estimated effects of modifiable factors varied by dementia risk, comparing both relative and absolute effects using interaction terms between dementia risk and each modifiable risk factor. Low dementia probability was associated with lower nursing home admission risk (RR=0.49; 95% CI: 0.41, 0.59). Physical inactivity (RR=1.27; 95% CI: 1.15, 1.41), ever smoking (RR=1.12; 95% CI: 1.01, 1.25), and not consuming alcohol (RR=1.28; 95% CI: 1.13, 1.45) predicted increased relative risk of nursing home admission regardless of cognitive status. The relative effects of modifiable risk factors were similar for those with low and high dementia risk. Although cognitive impairment associated with incipient dementia strongly predicts nursing home admission, this risk can be partially ameliorated with modifiable risk factors such as physical activity.