IntroductionDue to its proximity with the brain, cerebrospinal fluid (CSF) could be a medium of choice for the discovery of biomarkers of neurological and psychiatric diseases using untargeted analytical approaches.ObjectivesThis study explored the CSF lipidome in order to generate a robust mass spectral database using an untargeted lipidomic approach.MethodsCerebrospinal fluid samples from 45 individuals were analyzed by liquid chromatography coupled to high-resolution mass spectrometry method (LC-HRMS). A dedicated data processing workflow was implemented using XCMS software and adapted filters to select reliable features. In addition, an automatic annotation using an in silico lipid database and several MS/MS experiments were performed to identify CSF lipid species.ResultsUsing this complete workflow, 771 analytically relevant monoisotopic lipid species corresponding to 550 unique lipids which represent five major lipid families (i.e., free fatty acids, sphingolipids, glycerophospholipids, glycerolipids, and sterol lipids) were detected and annotated. In addition, MS/MS experiments enabled to improve the annotation of 304 lipid species. Thanks to LC-HRMS, it was possible to discriminate between isobaric and also isomeric lipid species; and interestingly, our study showed that isobaric ions represent about 50 % of the total annotated lipid species in the human CSF.ConclusionThis work provides an extensive LC/HRMS database of the human CSF lipidome which constitutes a relevant foundation for future studies aimed at finding biomarkers of neurological disorders.Electronic supplementary materialThe online version of this article (doi:10.1007/s11306-016-1023-8) contains supplementary material, which is available to authorized users.