Abstract: Objective：To explore altered lipid composition and complex metabolic networks associated with malignant differentiation of glioma. Methods：Shotgun lipidomics analyses in a series of human glioma cell lines with different tumorigenicity and metastasis potential, containing non-tumorigenic A172, tumorigenic U251, and metastatic U87MG. Results：A total of 1 541 unique lipids were identified originating from 3 major lipid categories-glycerophospholipids, glycerides, and sphingolipids via systematic lipidomics profiling. Both positive and negative ESI tandem mass spectrometry modes were used, and precursor ion (PI) scans and neutral loss (NL) scans were run respectively in each mode. GP species PC and PA had the increased trends in tumorigenic cell lines U251 and U87MG compared with A172, whereas remarkably decreased d18∶1-containing SP were observed with the increase of glioma tumorigenicity and metastatic potential (P＜0.05). Comparative lipidomics screening revealed 121 significantly changed lipids, and multivariate statistical analyses unraveled that altered lipid composition distinguished tumorigenic and metastatic cell lines U251 and U87MG from non-tumorigenic cell lines A172. Conclusions：Our lipidomics data shows obvious differences in lipid composition between non-tumorigenic and tumorigenic glioma cells. Differentially expressed lipids may be related to glioma metastasis and recurrence. These findings can lead to identify critical diagnostic biomarkers as well as developing therapeutic targets for glioma metastasis and recurrence.