Background: Physical activity (PA) may slow the development of dementia by reducing the accumulation of amyloid. Objective: We tested the hypothesis that higher levels of leisure-time PA in mid- or late-life were associated with lower brain amyloid burden in late-life among 326 non-demented participants from the Atherosclerosis Risk in Communities Study of brain florbetapir positron emission tomography (ARIC-PET) ancillary. Methods: Self-reported PA was quantified using a past-year recall, interviewer-administered questionnaire in mid-life (1987-1989, aged 45-64 years) and late-life (2011-2013, aged 67-89 years). Continuous PA estimates were classified as 1) any leisure-time PA participation (yes/no); 2) meeting the 2018 United States' PA guidelines (yes/no); and 3) per 1 standard deviation (SD) higher metabolic equivalent of task (MET) minutes per week (MET min wk-1). A brain magnetic resonance imaging scan with Florbetapir PET was performed in late-life. Adjusted odds ratios (OR) of elevated amyloid burden, defined as a global cortical standardized uptake value ratio (>1.2), compared to no elevated amyloid burden were estimated according to PA measures. Results: Among the 326 participants (mean age: 76 years, 42% male, 41% Black), 52% had elevated brain amyloid burden. Mid-life leisure-time PA did not show a statistically significant lower odds of elevated late-life amyloid burden (OR = 0.71, 95% CI: 0.43-1.18). A 1 SD (970 MET. min. wk-1) higher PA level in mid-life was also not significantly associated with elevated amyloid burden (OR = 0.89, 95% CI: 0.69-1.15). Similar estimates were observed for meeting versus not meeting PA guidelines in both mid- and late-life. Conclusion: Self-reported higher mid- and late-life leisure-time PA were not significantly associated with lower amyloid burden. Data show a trend of an association, which is, however, imprecise, suggesting replication in larger studies.