The aim of this study was to assess the lagged bidirectional associations between school connectedness and mental health problems in early adolescence. The longitudinal data used in this study were collected at five time points during a two-year period (each interval was a half year). The total sample for analysis consisted of 1015 adolescents aged 11.1-15.1 years from two junior high schools in Sichuan province, China. The School Connectedness Scale (SCS) and the Middle-School Student Mental Health Inventory (MMHI) were used five times. A cross-lagged structural equation model was used to assess the longitudinal associations. The correlation coefficients between the SCS and MMHI scores were - 0.173-0.364 (all "P" values < 0.05) with regard to five time points across two years. Negative lagged bidirectional associations were present between school connectedness and mental health problems both with a lag time of a half year and one year. There were different patterns of cross-lagged effects in the subgroups boys and girls and in the subgroups attending a school in an area that was heavily affected by the 2008 Wenchuan earthquake and the subgroups attending a school in an area that was not heavily affected by the earthquake. The results suggest that to promote mental health of adolescents in schools, attention should be given to strategies that aim to improve school connectedness, as well as to strategies to identify and manage mental health problems. Further observational studies are recommended in varied populations, as well as randomized intervention studies to evaluate the effects of school-based interventions.