Objective: Because leisure-time physical activity (LTPA) is protective against incident dementia, we hypothesized that LTPA is protective against decline in domain-specific cognitive performance.Methods: As part of the Northern Manhattan Study, LTPA was ascertained at enrollment using a validated in-person questionnaire. We assessed cognition in participants in the Northern Manhattan Study MRI substudy using a standard neuropsychological examination (NPE) (n 5 1,228), and a repeat examination was performed 5 years later (n 5 876). LTPA was summarized as the maximum intensity of any activity performed, classified as none to light intensity (physical inactivity) (90%) vs moderate to heavy intensity (10%). The NPE was subcategorized using standardized z scores over validated domains: processing speed, semantic memory, episodic memory, and executive function. We used multivariable linear regression models to examine the association of LTPA with initial and change in cognitive performance. Analyses were adjusted for sociodemographics, cardiovascular disease risk factors, and MRI findings (white matter hyperintensity volume, silent brain infarcts, cerebral volume).Results: No/low levels of LTPA were associated with worse executive function, semantic memory, and processing speed scores on the first NPE. The associations were slightly attenuated and no longer significant after adjusting for vascular risk factors. Cognitively unimpaired participants reporting no/low LTPA vs moderate/high levels declined more over With an expected increase in the mean age and proportion of the population older than 65 years, the public health burden of cognitive impairment and dementia will become substantial.1 There is substantial evidence for a contribution of subclinical cerebrovascular disease to dementia, both through direct vascular injury to the brain and/or modification of neurodegenerative processes. Increasing leisure-time physical activity (LTPA) is one target for dementia prevention because it is modifiable, does not require the use of medications, and provides substantial benefits for other diseases of aging.Prior studies have shown associations between LTPA and cognitive performance, 2-8 and a recent meta-analysis 9 has demonstrated a dose-response association between LTPA and subsequent risk of dementia. These studies tend to show a protective effect on only vascular dementia or on Alzheimer disease alone.10,11 Not all groups have documented an association between physical activity and cognition in older individuals. 5,12,13 Randomized clinical trials of LTPA *These authors contributed equally to this work.