To investigate the longitudinal predictive effects of peer victimization, loneliness, and depression, a cross lag analysis was conducted on 785 junior high school students at three-month intervals. The results showed that: (1) there was a bidirectional predictive relationship between peer victimization and depression, loneliness, and both depression and loneliness at T1 and T2 time points. (2) Loneliness T2 partially mediated the effect of peer victimization T1 on depression T2, and loneliness T2 also partially mediated the effect of depression T1 on peer victimization T2. The results indicate that there is a cyclical relationship between peer victimization and internalization problems among middle school students. Educators can prevent peer victimization and improve the mental health level of middle school students from the perspective of reducing loneliness.