Aim Through the analysis of serum metabolites in coronary heart disease (CHD) patients with different degrees of coronary artery lesion, to find new biomarkers for predicting and diagnosing CHD. Methods 92 CHD patients who met the experimental requirements from July 2017 to December 2018 in Tianjin Third Central Hospital were selected for serum analysis, including 57 males and 35 females. Among them, 24 patients had coronary artery stenosis ＜50%, 19 patients had at least one coronary artery stenosis ≥50% and ≤70%, 23 patients had at least one coronary artery stenosis ＞70%, 26 patients had acute coronary occlusion. Liquid chromatography-mass spectrometry was used to detect the chromatographic peak and mass spectrum peak of metabolites in samples, in order to determine the metabolites.The clinical diagnostic efficacy of characteristic metabolites was evaluated by ROC curve. Results 47 different metabolites were identified in metabolic disorders, and 10 characteristic metabolites were finally identified. With the increase of the degree of coronary artery stenosis, the metabolic pathway changed, including the decrease of glycerol phospholipid and fatty acid metabolism, and the increase of glycosphingolipid and purine metabolism. ROC curve analysis showed that the area under the curve of characteristic metabolites was all greater than 0.75. Conclusion Metabolic disorders are present in CHD patients with different degrees of coronary artery lesion, and the characteristic metabolites have diagnostic value.