Hypertension continues to pose a huge burden to global public health. Abnormal metabolism not only serves as a risk factor for hypertension but also acts as a driving force in its aggravation. However, there remains a lack of large-scale causal demonstration based on extensive samples. Our study aims to investigate the causal relationship between metabolism and primary hypertension (PH) using Mendelian randomization analysis. We used genome-wide association studies instrumental variables for Mendelian randomization association analysis integrating the diagnosis results of PH in 3 populations from East Asia, the Middle East, and Africa with serum metabolites and metabolite ratios. This allowed us to identify predictive metabolites and metabolic pathways for diagnosing or treating PH. Inverse-variance weighting was the main model for establishing causal associations. In addition horizontal pleiotropy test, linkage disequilibrium test, and sensitivity analysis were employed to test the explanatory power of instrumental variables. A total of 10,922 cases of PH and 8299 cases of metabolomics detection cohorts were included in the study. In East Asian, Middle Eastern, and African populations, we found 36, 57, and 40 known metabolites respectively strongly associated with PH (P <
.05). Cross-section and meta-analysis of these strongly correlated metabolites across the 3 ethnic groups revealed 7 common metabolites. Notably, elevated isoleucine (odds ratio = 0.74, 95% confidence interval: 0.56-0.96) was demonstrated as a potential protective factor against PH across 3 ethnic groups. The metabolites associated with PH have certain polymorphisms in different populations. Isoleucine may be a promising biomarker for PH diagnosis or treatment, but more clinical validation is needed.