Physical activity is associated with numerous health benefits in youth; however, these benefits could extend further than health, into education. Our aim was to systematically review and combine in meta-analyses evidence concerning the association between physical activity and the dimensions of school engagement, including behavior (e.g., time-on-task), emotions (e.g., lesson enjoyment), and cognition (e.g., self-regulated learning). We conducted meta-analyses using structural equation modeling on results from 38 studies. Overall, physical activity had a small, positive association with school engagement (d = 0.28, I[superscript 2] = 0.86), 95% confidence interval [0.12, 0.46]. This association was moderated by study design, with significant associations shown in randomized controlled trials but not in studies employing other designs. Risk of bias was also a significant effect moderator, as studies with a low risk of bias showed significant associations but not high risk of bias studies. Altogether, these results suggest that physical activity could improve school engagement.