To better understand the longitudinal relations between risk and relationship variables during early to middle adolescence, we employed latent variable cross-lagged panel models using data (N = 1,050) from the National Institute of Child Health and Human Development Study. Results showed self-reported individual risk behavior ([beta] = 0.77-0.88) and mother-reported child-parent relationship quality (CPRQ; [beta] = 0.59-0.76) were stable, whereas the self-report of one's peer's risk behavior ([beta] = 0.21-0.27) was less stable. After controlling for previous levels of peer risk behavior and CPRQ, higher levels of individual risk behavior resulted in subsequent increases in the number of peers who demonstrated risk behavior ([beta] = 0.38-0.54) across early to middle adolescence. CPRQ and peer risk behavior, however, did not influence subsequent individual risk behavior. School psychologists can encourage friendship selection strategies that reduce potential risk affiliation while increasing healthy, supportive friendships and teach adolescents stress management skills.