Prior studies have indicated that prosocial behavior might be a protective factor for developing internalizing and externalizing behavioral problems. However, little research has been conducted on within-person changes of prosocial behavior and behavioral problems over time. With random-intercept cross-lagged panel models (RI-CLPMs), the current study analyzed longitudinal associations between prosocial behavior and behavioral problems in two twin cohorts (98% Western European): in early childhood (age M = 4.77, SD = .58, 52% girls, N = 440) and middle childhood (age M = 7.94, SD = .67, 51% girls, N = 512). To obtain robust results, two parental reported questionnaires and an observational task were used as prosocial behavior assessments. In line with the literature, we found a significant between-person association between externalizing behavior and parent reported prosocial behavior in middle childhood, but not in early childhood. Some evidence indicated that changes in externalizing problems affect later prosocial behavior in middle childhood. Overall, however, the RI-CLPMs provided most support for the hypothesis that within-person changes in prosocial behavior are not related to within-person changes in behavioral problems. Thus, our findings did not support the hypothesis that increased prosocial behavior directly results in decreased behavioral problems, but emphasizes the need to take into account the multifaceted nature of prosocial behavior.