Methods for rapid and reliable microbial identification are essential in modern healthcare. The ability to detect and correctly identify pathogenic species and their resistance phenotype is necessary for accurate diagnosis and efficient treatment of infectious diseases. Bottom-up tandem mass spectrometry (MS) proteomics enables rapid characterization of large parts of the expressed genes of microorganisms. The generated data is however highly fragmented, making down-stream analyses complex. Here we present TCUP, a new computational method for typing and characterizing bacteria using proteomics data from bottom-up tandem MS. TCUP compares the generated protein sequence data to reference databases and automatically finds peptides suitable for characterization of taxonomic composition and identification of expressed antimicrobial resistance genes. TCUP was evaluated using four clinically relevant bacterial species (Escherichia coli, Pseudomonas aeruginosa, Staphylococcus aureus and Streptococcus pneumoniae), using both simulated data generated by in silico peptide digestion and experimental proteomics data generated by liquid chromatography-tandem mass spectrometry (MS/MS). The results showed that TCUP performs correct peptide classifications at rates between 90.3% and 98.5% at the species level. The method was also able to accurately estimate the relative abundances of individual species in mixed cultures. Furthermore, TCUP could identify expressed beta-lactamases in an ESBL-producing E. coli strain, even when the strain was cultivated in the absence of antibiotics on non-selective media. Finally, TCUP is computationally efficient, easy to integrate in existing bioinformatics workflows and freely available under an open source license for both Windows and Linux environments.