Updated project metadata. The current technique used for microbial identification in hospitals is MALDI-TOF MS. However, it suffers from important limitations, in particular for closely-related species or when the database used for the identification lacks the appropriate reference. In this work, we set up a high throughput LC-MS/MS top-down proteomics platform dedicated to intact bacterial protein analysis. Using Escherichia coli as a model, all steps of the workflow were optimized: protein extraction, on-line liquid chromatographic separation, MS/MS fragmentation and data analysis. Using the optimized parameters, about 220 proteins, corresponding to more than 500 proteoforms, could be identified in a single run. We then demonstrated the suitability of the developed platform for the characterization and discrimination of enterobacterial pathogens rather undistinguishable by MALDI-TOF although leading to very different clinical outcomes. For each pathogen, we identified specific proteoforms that could potentially be used as biomarkers. We also improved the characterization of poorly described bacterial strains. Our results highlight the advantage of targeting proteoforms vs peptides for accurate bacterial characterization, and qualify top-down proteomics as a promising tool in clinical microbiology.