Objective To explore potential serum biomarkers of children with Kashin-Beck Disease (KBD) and the metabolic pathways to which the biomarkers belong. Methods A two-stage metabolomic study was employed. The discovery cohort included 56 patients, 51 internal controls, and 50 external controls. The metabolites were determined by HPLC-(Q-TOF)-MS and confirmed by Human Metabolome Databases (HMDB) and Metlin databases. MetaboAnalyst 3.0 and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database were used to analyze the metabolic pathways of the candidate metabolites. The use of HPLC-(Q-TRAP)-MS enabled quantitative detection of the target metabolites which were chosen using the discovery study and verified in another independent verification cohort of 31 patients, 41 internal controls, and 50 external controls. Results Eight candidate metabolites were identified out in the discovery study, namely kynurenic acid, N-α-acetylarginine, 6-hydroxymelatonin, sphinganine, ceramide, sphingosine-1P, spermidine, and glycine. These metabolites exist in sphingolipid, glutathione, and tryptophan metabolic pathways. In the second-stage study, five candidate metabolites were validated, including kynurenic acid, N-α-acetylarginine, sphinganine, spermidine, and sphingosine-1P. Except for spermidine, all substances exhibited low expression in the case group compared with the external control group, and the difference in levels of sphinganine, spermidine, and sphingosine-1P was statistically significant. Conclusion The direction of change of levels of sphinganine, spermidine, and sphingosine-1P in the two-stage study cohorts was completely consistent, and the differences were statistically significant. Therefore, these substances can be used as potential biomarkers of KBD. Furthermore, these results raise the possibility that sphingolipid metabolic pathways may be closely related to KBD.