With the popularity of online games, there are more and more violent elements in video games. Previous studies have mainly examined the impact of violent video game exposure on individuals’ traditional aggressive behaviors in the real world, but few have explored the relationship between the former and individuals’ online aggressive behaviors. This study explored the influencing factors and the forming mechanism of online aggressive behaviors among college students, so as to provide empirical reference and targeted suggestions for preventing and intervening college students’ online aggressive behaviors. This study adopted a cross-sectional design. A total sample of 946 students from several universities in China was selected, with 529 males (55.92%) and 417 females (44.08%). The age of the subjects ranged from 17 to 26, with an average age of 20.74±1.61 years old. The questionnaires included the Video Game Questionnaire (VGQ), Adolescent Online Aggressive Behavior Scale (AOABS), VIO sub-scale of Abnormal Personality Risk Factors Inventory (FUMM-APRFI), and the Trait Anger Scale (TAS). The results showed that: (1) violent video game exposure (VVGE) positively predicted college students’ online aggressive behaviors; (2) trait anger moderated the relationship between VVGE and college students’ online aggressive behaviors; to be specific, at high levels of trait angry, VVGE positively predicted college students’ online aggressive behaviors (βsimple =.40, p<.001, with the confidence interval of 95% [.79, .16]), while at low levels of trait anger, the effect of VVGE on college students’ online aggressive behaviors was not significant (βsimple =.17，p>.05, with the confidence interval of 95% [.46, -.14]); (3) trait anger moderated the relationship between VVGE and college students’ attitudes toward violence; to be specific, at high levels of trait angry, VVGE positively predicted college students’ attitudes toward violence (βsimple =.37，p<.001, with the confidence interval of 95% [.75, .13]), while at low levels of trait anger, the effect of VVGE on college students’ attitudes toward violence was not significant (βsimple =.14，p>.05, with the confidence interval of 95% [.35, -.08]); (4) attitudes toward violence not only partially mediated the effect of VVGE on college students’ online aggressive behaviors, but also partially mediated the effect of interaction between trait anger and VVGE on college students’ online aggressive behaviors. It is concluded that attitudes toward violence acted as a partial mediator in the process of VVGE influencing college students’ online aggressive behaviors; meanwhile, trait anger moderated the effects of VVGE on college students’ attitudes toward violence and online aggressive behaviors. This study extends the aggressive behaviors from the real world to the virtual network, and further enriches the empirical evidence of the General Aggressive Model. The findings provide evidence for the mechanism of trait anger and attitudes toward violence in the process of violent video game exposure affecting online aggressive behavior, and help to provide a new perspective for the intervention of college students’ online aggressive behaviors.