Abstract Background The joint impact of sedentary behavior and physical activity on obesity has not been assessed in a large cohort followed from adolescence to adulthood. Methods Nationally representative longitudinal data from Waves II (1995; mean age: 15.9) and III (2001; mean age: 21.4) of the National Longitudinal Study of Adolescent Health (n = 9,155) were collected. Sex-stratified multivariate logistic regression analysis assessed the odds of obesity associated with Wave II MVPA and screen time, controlling for sociodemographic characteristics and change in MVPA and screen time from Wave II to III. Obesity was defined using body mass index (BMI, kg/m2) International Obesity Task Force cut-points at Wave II and adult cut-points at Wave III (BMI ≥ 30). Results In males, adjusted odds of prevalent obesity was strongly predicted by MVPA bouts [OR (95% CI): OR6 vs. 1 MVPA bouts = 0.50 (0.40, 0.62); OR4 vs. 40 hrs screen time = 0.83 (0.69, 1.00)]. In females, greater MVPA bouts and lower screen time correlated with lower prevalent obesity [OR (95% CI): OR6 vs. 1 MVPA bouts = 0.67 (0.49, 0.91); OR4 vs. 40 hrs screen time = 0.67 (0.53, 0.85)]. Longitudinally, adolescent screen time hours had a stronger influence on incident obesity in females [OR (95% CI): OR4 vs. 40 hrs = 0.58 (0.43, 0.80)] than males [OR (95% CI): OR4 vs. 40 hrs = 0.78 (0.61, 0.99)]. Longitudinal activity patterns were not predictive of incident obesity. Conclusion Reducing screen time during adolescence and into adulthood may be a promising strategy for reducing obesity incidence, especially in females.