Although physical activity and sleep may influence each other, little is known about the bidirectional association of the two behaviors. Analyses included 38,601 UK Biobank participants (51% female, 55.7 ± 7.6 years old, 6.9 ± 2.2 years of follow-up). Physical activity was categorized by the weekly metabolic equivalent of task minutes (highly active: ≥ 1200; active: 600 to <1200; inactive: < 600), and sleep patterns were determined using a composite score of healthy sleep characteristics: morning chronotype, adequate sleep duration (7-8 h/d), never or rare insomnia, never or rare snoring, and infrequent daytime sleepiness. We categorized the sleep score into three patterns (healthy: ≥ 4; intermediate: 2-3; poor: ≤ 1). Multiple logistic regressions examined the association of baseline (or the temporal changes in) sleep/physical activity with physical inactivity/poor sleep at follow-up. Participants with an intermediate or poor sleep pattern at baseline had higher odds (adjusted odds ratio: 1.24 [1.17, 1.32] and 1.65 [1.45, 1.88], respectively) for physical inactivity at follow-up, compared to those with healthy sleep, while shifting to a healthy sleep pattern over time attenuated these adverse associations. Compared to individuals highly active at both time points, being physically inactive at baseline and reducing physical activity over time were both associated with higher odds for poor sleep at follow-up. In conclusion, sleep improvements over time benefitted physical activity at follow-up, while reduced physical activity had a detrimental effect on sleep patterns at follow-up. Our results provide scope for interventions to concurrently target physical activity and sleep.Copyright © 2020 American Health Foundation. Published by Elsevier Inc. All rights reserved.