Abstract Background The rupture of carotid atherosclerotic plaque, primarily driven by plaque components, is among the leading causes of ischemic stroke. Although physical activity (PA) stands as a major modifiable target for prevention of atherosclerotic diseases, recent studies suggested a potential adverse impact of high-volume PA on atherosclerotic plaque burden in athletes. Data remains scarce from the community-dwelling population regarding to the impact of PA on the longitudinal change of carotid plaque components. Objectives To assess the association between PA and the incidence of carotid plaque components and explore whether the association of PA with stroke and all-cause mortality differs by the presence of plaque components. Methods Between 2007 and 2012, in a community- dwelling cohort in Netherlands, 1327 stroke-free persons (mean ± standard deviation age 72 ± 9 years, 45% women) with carotid atherosclerosis (carotid intima-media thickness > 2.5 mm) were invited for magnetic resonance imaging (MRI) to assess carotid plaque components. From these, 696 persons (1267 carotid arteries) underwent a follow-up MRI between 2014 and 2017 (median between-scan interval 6 years). Carotid plaque components included calcification, intraplaque hemorrhage (IPH), and lipid-rich necrotic core (LRNC). The volume of moderate, vigorous and total PA (unit: metabolic equivalent task hour, MET-h) was assessed using LASA Physical Activity Questionnaires around the time of the first MRI scan. Participants were followed for the occurrence of stroke or all-cause mortality until December 31, 2019. Generalized estimation equation and Cox regression models were used in the analyses, with adjustments for plaque size and other cardiovascular risk factors. Results Compared to lowest quintile PA group, participants in the highest quintile total PA group had a higher risk of new onset of plaque components, odds ratio (95%CI) 2.69 (1.21; 5.94) for calcification, 1.35 (0.82; 2.24) for LRNC and 1.93 (1.12; 3.34) for IPH (Figure 1). Similar associations were observed for vigorous PA but not for moderate PA. Over a median follow-up of 9.5 years, 118 persons suffered a stroke and 298 died. As shown in table 1, baseline IPH, rather than LRNC or calcification, modulated the associations of PA with stroke and all-cause mortality (P-interaction <0.05): higher total PA was associated with a lower risk of stroke among IPH-free persons, with the hazard ratio (95% CI) of 0.929 (0.867; 0.996) per 10 MET-h/week increase; no associations between PA and stroke were found among those with IPH. Similar associations were found between vigorous PA and all-cause mortality. Conclusions Among the middle-aged and elderly with subclinical atherosclerosis, high level PA, especially high vigorous PA, may lead to an unstable carotid plaque. The beneficial effect of PA on stroke and all-cause mortality is only among persons with a stable carotid plaque profile, especially without IPH.