Background <p>Bedtime procrastination has emerged as a pervasive health risk behavior among young adults, whereas physical activity represents a fundamental health-promoting behavior. Although the strength model of self-control theoretically suggests that engaging in physical activity may replenish the self-regulatory resources necessary to resist evening procrastination, empirical evidence remains inconclusive. Previous cross-sectional and traditional cross-lagged studies have consistently confounded stable, trait-like differences between individuals with dynamic, state-like fluctuations within individuals. This study aimed to disentangle these effects to investigate the dynamic longitudinal interaction among physical activity, self-control, and bedtime procrastination.</p> Methods <p>A three-wave longitudinal survey design with one-month intervals was employed. A final matched sample of 385 undergraduate students (mean age = 20.45 years; 53.8% female) completed validated self-report measures assessing physical activity, self-control, and bedtime procrastination across all three time points. To isolate intra-individual temporal dynamics from time-invariant baseline characteristics, the data were analyzed utilizing a random intercept cross-lagged panel model (RI-CLPM), controlling for chronotype and time-varying perceived stress.</p> Results <p>At the between-person (trait) level, the random intercepts revealed that average physical activity was significantly and positively correlated with average self-control (<i>r</i> = .41, <i>p</i> &lt; .001) and negatively correlated with average bedtime procrastination (<i>r</i> = − .35, <i>p</i> &lt; .001). Crucially, at the within-person (state) level, a significant temporal cascade emerged. An intra-individual increase in physical activity at Time 1 was prospectively associated with an increase in self-control at Time 2 (<i>β</i> = 0.18, <i>p</i> &lt; .01), which subsequently predicted a decrease in bedtime procrastination at Time 3 (<i>β</i> = −0.24, <i>p</i> &lt; .001). Monte Carlo simulations confirmed a significant within-person longitudinal mediation effect (95% CI [− 0.082, − 0.015]).</p> Conclusion <p>The findings provide prospective longitudinal evidence that health-promoting and health risk behaviors covary dynamically over time in association with shared cognitive resources. State-level increases in physical activity were prospectively associated with subsequent reductions in bedtime procrastination, potentially through enhanced self-control capacity. These results highlight the practical utility of cross-domain interventions, suggesting that daytime physical exertion may serve as a potential target for interventions aiming to reduce nocturnal self-regulatory failures. These conclusions should be interpreted with caution, given the observational nature of the design.</p>