A rising global concern, adolescent aggression has been linked to adolescents' mental health conditions, and vice versa. Although longitudinal relationships between the two have been studied, within-person associations between these variables, which are important for informing interventions, have not been adequately examined. To bridge that research gap, this study examined the within-person associations between aggression (i.e., reactive, proactive, and cyber aggression) and mental health problems (i.e., depressive and anxious symptoms), as informed by the frustration-aggression theory and the failure model. Three-wave longitudinal data were collected from a sample of Chinese adolescents (N = 1422; 50.9% girls; mean age = 13.56 years) at three time points, each separated by one-year intervals. The data were analyzed using a random-intercept cross-lagged panel model (RI-CLPM), revealing several within-person relationships. The presence of symptoms of depression and anxiety at T2 predicted increased cyber aggression at T3, and depressive symptoms at T2 also predicted an increase in reactive aggression at T3 (p < 0.1). In addition, proactive aggression at T2 predicted an increase in depressive symptoms at T3 (p < 0.1), and reactive aggression at T1 predicted a reduction in symptoms of anxiety at T2. All aggression- and mental health-related variables were significantly correlated at the between-person level. Moreover, the results of the multiple-group RI-CLPMs showed that gender influenced the relationships between proactive aggression and symptoms of depression and anxiety. The study's results lend partial support to the notion of bidirectional relationships between adolescent aggressive behaviors and mental health conditions, as well as to the frustration-aggression theory and the failure model. Insights into the interactions between adolescents' mental health problems and aggression can inform prevention and intervention strategies.