Myelodysplastic syndrome (MDS) is a neoplastic disease originating from hematopoietic stem cells. Currently, hematopoietic stem cell transplantation (HSCT) is the most effective cure, although lenalidomide, azacytidine, and decitabine have been applied to relieve symptoms of MDS. The purpose of this study was to evaluate the changes in endogenous metabolites by applying a UHPLC–MS (ultra–high‐performance liquid chromatography–MS) metabolomics approach and to investigate metabolic pathways related to MDS. An untargeted metabolomics approach based on UHPLC–MS in combination with multivariate data analysis, including partial least squares discrimination analysis and orthogonal partial least squares discriminant analysis, was established to investigate potential biomarkers in the plasma of MDS patients. As a result, 29 biomarkers were identified to distinguish between MDS patients, HSCT patients, and healthy controls, which were mainly related to inflammation regulation, amino acid metabolism, fatty acid metabolism, and energy metabolism. To our knowledge, this is the first time where plasma metabolomics was combined with HSCT to study the pathogenesis and therapeutic target of MDS. The identification of biomarkers and analysis of metabolic pathways could offer the possibility of discovering new therapeutic targets for MDS in the future.