Objective: The cross lagged analysis model is used to analyze the causal relationship between physical health education and adolescent health literacy, so as to provide empirical references for physical health education and adolescent health literacy improvement in China. Methods: Cluster sampling method was used to select 24 classes of grade 7 students from four junior high schools in Xiangyang City,Hubei Province,Dalian City,Liaoning Province and Liaoyuan City,Jilin Province.The longitudinal follow up design with an interval of 18 weeks was administered among 696 junior middle school students. The Revised School Physical Health Education Scale and Junior Middle School Students Health Literacy Scale were used. Results: T1 exercise supervision could significantly and positively predict T2 health literacy( β=0.18 ), T1 health literacy could also significantly and positively predict T2 exercise supervision( β=0.18)(P <0.01). T1 health education could significantly and positively predict T2 health literacy( β=0.57, P <0.01), T1 health literacy could not significantly and positively predict T2 health education( β=0.03, P >0.05). T1 physical education teaching could significantly and positively predict T2 health literacy( β =0.39), T1 health literacy could also positively and significantly predict T2 physical education teaching( β=0.10)(P <0.05). The three dimensions of physical health education had positive predictive effect on the three dimensions of health literacy( P <0.05). Conclusion In the process of physical health education, health education is the main factor affecting teenagers health literacy, with a vertical causal relationship between the two variables. There is a bi directional positive relationship between sports supervision, physical education teaching and adolescent health literacy.