AIM: To explore the causal relationship between several possible behavioral factors and high myopia (HM) using multivariable Mendelian randomization (MVMR) approach and to find the mediators among them with mediation analysis. METHODS: The causal effects of several behavioral factors, including screen time, education time, time spent outdoors, and physical activity, on the risk of HM using univariable Mendelian randomization (MR) and MVMR analyses were first assessed. Genome-wide association study summary statistics of serum metabolites were also used in mediation analysis to determine the extent to which serum metabolites mediate the effects of behavioral factors on HM. RESULTS: MR analyses indicated that both increased time spent outdoors and a higher frequency of moderate physical activity significantly reduced the risk of HM. Further MVMR analysis confirmed that moderate physical activity independently contributed to a lower risk of HM. Additionally, MR analyses identified 13 serum metabolites significantly associated with HM, of which 12 were lipids and one was an amino acid derivative. Mediation analysis revealed that six lipid metabolites mediated the protective effects of moderate physical activity on HM, with the highest mediation proportion observed for 1-(1-enyl-palmitoyl)-GPC (p-16:0; 30.83%). CONCLUSION: This study suggests that in addition to outdoor time, moderate physical activity habits may have an independent protective effect against HM and pointed to lipid metabolites as priority targets for the prevention due to low physical activity. These results emphasize the importance of physical activity and metabolic health in HM and underscore the need for further study of these complex associations.