Objective: To explore the longitudinal relationship between peer attachment, peer trust and loneliness, and to provide reference for the effective adolescent mental health promotion. Methods: A convenient sampling method was used to select 1 013 first year senior high school students from 2 high schools in Guizhou Province and Shandong Province. A longitudinal design was adopted. The Revised Experiences in Close Relationships relationship Structures Scale(ECR-RS), Trust Scale and University of California at Los Angels Loneliness Scale were administered in Nov. 2020, Dec. 2021(T1), as well as in Jan. 2021 and Jan. 2022(T2). Results: Peer trust at two time points was negatively correlated with attachment anxiety, attachment avoidance and loneliness( r=-0.50--0.17, P <0.01), while attachment anxiety, avoidance and loneliness were positively correlated( r=0.11- 0.41 , P <0.01). T1 attachment anxiety significantly predicted T2 loneliness( β=0.16, P <0.01), and T1 loneliness significantly predicted T2 attachment anxiety and avoidance( β=0.19, 0.15, P <0.01). Correlation between stability of loneliness was higher than attachment anxiety( CR=7.12, P <0.01). Correlation between stability of peer trust was higher than attachment avoidance( CR=2.40, P <0.01). Conclusion Loneliness affects attachment avoidance and peer trust unidirectionally. There is mutual influence between loneliness and attachment anxiety, with larger impact from loneliness. Intervention aiming for attachment promotion might consider loneliness reduction.