Experimental studies suggest that physical activity or exercise can prevent or delay the onset of age-related cognitive impairment or dementia. Several epidemiologic studies have attempted to address this issue by using widely varying definitions of physical activity as well as differing methods to assess cognitive function or dementia. Despite the variability in study design, longitudinal studies report that the risk of dementia, cognitive impairment, cognitive decline, and Alzheimer's disease is lower among persons engaging in high levels of physical activity, relative to those engaging in low levels of physical activity. There is also evidence of a dose-response relationship between physical activity and preservation of cognitive function, even though small increases in physical activity from a sedentary state can reap comparatively large incremental benefits. Reductions in the relative risk of vascular dementia associated with physical activity were weaker and more variable. Although many studies had a follow-up duration of around 5 years, at least one study has found evidence that exercise in midlife reduces the risk of dementia in late life. How interactions between physical activity and the APOE e4 allele affect the risk of cognitive impairment is inconsistent. The available epidemiologic studies support the concept that physical activity might prevent or delay the onset of cognitive decline. Whether that effect will translate into a reduction in the prevalence of dementia is unknown.