PURPOSE: To investigate potential bidirectional relationships between sport participation and mental health during early adolescence. METHODS: Data were taken from wave 5 (2012) and wave 6 (2014) of the K-Cohort of the Longitudinal Study of Australian Children. In total, there were 4023 participants aged 12.41 (SD = 0.49) years at baseline, and this sample were followed up 24 months later. Cross lagged panel models were used to examine bidirectional relationships between sport participation (hours.wk for team, individual and total sport participation) and mental health (total psychological difficulties, internalising problems, and externalising problems) as measured by the parent-report version of the Strengths and Difficulties Questionnaire. RESULTS: Bidirectional relationships were evident between time involved in sport and overall mental health (Sport12–>SDQ14: beta = -.048; SDQ12–>Sport14: beta = -.062). Bidirectional relationships were also evident between time involved in sport participation and internalising (social and emotional) problems (Sport12–>SDQ14: beta = -.068; SDQ12–>Sport14: beta = -.067). The relationship between time in organised sport and externalising problems (conduct problems and inattention/hyperactivity problems) was not bidirectional. Externalising problems predicted later sports participation (beta = -.039), but not vice versa. CONCLUSION: Findings demonstrate bidirectional relationships between sport participation and adolescent mental health. The design and implementation of youth sport programs should maximise mental health benefits, and programs should be designed, implemented, and marketed to be attractive to participants with poor psychosocial health.