<p>Cyber aggression is a developmental concern during adolescence. Grounded in the General Aggression Model, this three-wave longitudinal study examined whether violence exposure was longitudinally associated with cyber aggression and whether positive rumination buffered these associations among Chinese adolescents. Participants included 3,108 students from 16 middle schools in Jiangxi Province, China, at bLongitudinal Moderating Role of Positive Rumination Across Genderaseline (54.9% girls; mean age = 14.41 years), of whom 2,904 completed all three waves across 6-month intervals. Latent growth models with full information maximum likelihood estimation were used to estimate changes in violence exposure, positive rumination, cyber relational aggression (CRA), and cyber overt aggression (COA). Violence exposure was positively associated with both the initial levels and slopes of CRA (<i>β</i> = 0.28 and 0.20) and COA (<i>β</i> = 0.23 and 0.16). Positive rumination exhibited a selective buffering effect: for CRA, both the intercept-level and slope-level interactions were negative (<i>β</i> = -0.13 and − 0.09), whereas for COA, only the intercept-level interaction was significant (<i>β</i> = -0.10). Girls exhibited stronger associations between violence exposure and CRA, and evidence of a slope-level buffering effect of positive rumination was clearer within the CRA model for girls. In contrast, boys showed a stronger association between violence exposure and the initial level of COA. Overall, violence exposure was longitudinally linked to cyber aggression, and positive rumination may serve as a partial protective factor, particularly for relational forms of cyber aggression. However, given the observational design, these findings should be interpreted as longitudinal associations rather than definitive causal effects.</p>