Although adolescents around the world play video games, little is known about their longitudinal effects on adolescents from Eastern countries. This large longitudinal violent video game study has 4 strengths. First, it is the first longitudinal study conducted with Chinese adolescents. Second, it examines moral disengagement as a possible mediating variable. Because violent video games reward immoral behaviors (e.g., stealing cars, killing characters), they might lead players to believe immoral behaviors are "no big deal." These moral disengagement beliefs might "bleed over" from the virtual world to the real world after the game is turned off, by increasing subsequent aggression. Third, it tests the hypothesis that violent video game effects are stronger for early adolescents than for late adolescents, and for males than for females. Fourth, it separates between-person effects from within-person effects. The theoretical foundation for the study is Social Cognitive Theory, which proposes that 3 interacting components determine what events result in observational learning: (a) context (i.e., violent video game exposure), (b) personal cognition (i.e., moral disengagement), and (c) behavioral outcome (i.e., aggression). Participants in this 3-wave (each wave was 6 months apart) longitudinal study were 1,340 Chinese adolescents (45.1% boys) 12- to 19-years-old (M = 14.86, SD = 1.50). Moral disengagement was a significant mediator of the longitudinal relationship between violent game exposure and aggression in the cross-lagged panel and between-person analyses, but not in the within-person analysis. As expected, effects were stronger for early adolescents than for late adolescents, especially between persons. However, gender did not moderate the effects.