The relationship between paternal/maternal rejection and adolescent risk-taking involves a bidirectional interaction, yet few studies have examined the adolescent-driven and parent-driven effects within parent-child interactions from a dynamic developmental perspective. Moreover, existing research has focused solely on one parent (particularly mothers), neglecting the role of fathers or failing to distinguish differences between mothers and fathers in their interactions with children. The present study used cross-lagged panel model and latent growth mixture model to examine the dynamic relationship and mechanism between paternal/maternal rejection and adolescent risk-taking behaviors. A total of 2024 adolescents (47.5% girls; Mage at Wave 1 = 13.07 years) were followed for three times with six-month intervals. The results showed that: (1) Parental roles exhibited significant differences in the bidirectional relationship between parental rejection and adolescent risk-taking behavior. Specifically, the paternal-driven effect was significantly stronger than the maternal-driven effect; Compared to paternal rejection, the adolescent-driven effect on maternal rejection was stronger, which was also significantly stronger than the maternal-driven effect. (2) Parental rejection and adolescent risk-taking exhibited distinct heterogeneous developmental trajectories. Parental rejection: paternal increase and maternal stable rejection group (5.14%), the parental rejection growth group (8.60%), the parental rejection low stability group (81.18%) and the maternal increase parental stable rejection group (5.09%). Adolescent risk-taking behavior: the low risk-taking growth group (13.19%), the low risk-taking stability group (83.60%), and the high risk-taking volatility group (3.21%). (3) Adolescents' self-disclosure to fathers mediated the effect of paternal rejection on adolescent risk-taking behavior, while adolescents' self-disclosure to mothers mediated the relationship between maternal rejection and adolescent risk-taking behavior in a bidirectional manner. There were gender differences in the mediation models. The findings have practical implications for fostering effective parenting practices and nurturing healthy behavioral patterns among adolescents.