Backgrounds and aimsIn toxicology, LC-HRMS for untargeted screening yields a great deal of high quality spectral data. However, there we lack tools to visualize/organize the MS data. We applied molecular networking (MN) to untargeted screening interpretation. Our aims were to compare theoretical MS libraries obtained in silico with our experimental dataset in patients to broaden its application, and to use the MetWork web application for metabolite identification. MethodsSamples were analyzed using an LC-HRMS system. For MN, data was generated using MZmine, and analyzed and visualized using MetGem. MetWork annotations were filtered and this file was used for annotation of the previously obtained MN. Results155 compounds including drugs found in patients were recorded. Using this dataset, we confirmed in 60 patients intake of tramadol, amitriptyline bromazepam, and cocaine. The results obtained by the reference methods were confirmed by MN approaches.. Eighty percent of the compounds were common to both conventional and MN approaches. Using MetWork, metabolites and parent drugs such as amitriptyline, its metabolite nortriptyline and amitriptyline glucuronide phase 2 metabolites were anticipated and proposed as putative annotations. ConclusionThe workflow increases confidence in toxicological screening by highlighting putative structures in biological matrices in combination with CFM-ID (Competitive Fragmentation Modeling for Metabolite Identification) and MetWork to extend the annotation of potential drugs even without a reference standard.