The residue determination of the widely used herbicide glyphosate (GLY) is highly problematic due to its amphoteric character, low mass and lack of chemical groups that might facilitate its detection. Most methods developed up to now have employed pre-column or post-column derivatization to form fluorescent derivatives and/or to reduce the polar character of the analyte facilitating its chromatographic retention. The aim of this work is to evaluate the feasibility of performing the direct LC–MS/MS determination of GLY residues in vegetables. After testing several Hydrophilic Interaction Liquid Chromatography (HILIC) columns, Obelisc N was selected due to its better chromatographic retention. LC–MS/MS determination has been performed in negative ionization mode, monitoring up to four transitions to give high reliability to the identification/confirmation process. This approach has been evaluated for the determination of GLY residues in rice, maize and soybean samples, and the method validated at different concentrations in compliance with the maximum residue limits established in the current legislation. After sample extraction with water, a combination of extract dilution, partition with dichloromethane, and solid phase extraction (SPE) using Oasis HLB cartridges (depending on the sample matrix under analysis) was applied. Quantification was made by using isotope-labeled GLY as internal standard and calibration in solvent. The methodology developed allows the rapid determination of GLY residues avoiding the derivatization step typically applied for this herbicide. The most critical issue is the robustness of the Obelisc N column, which was found to suffer rapid degradation with time. Extreme care and continuous testing of retention times and peak shapes is required for a reliable determination.