We report a novel, peptide identification-free shotgun proteomics workflow as a bacterial fingerprinting method. This method uses a similarity-clustering algorithm to segregate the mass spectra that are presumably derived from individual peptide speciess and merge them into discrete units of consensus spectra that comprise the proteomic fingerprints of the bacterial isolates being investigated. The novel method was compared to a traditional peptide identification-based shotgun proteomics workflow and a commonly used PCR-DNA fingerprinting technique for performance benchmarking in differentiating 73 isolates of E. coli by their animal sources (human, cow, dog, and pig). The fingerprints generated using by the novel method were richer in information, more discriminative in separating the E. coli isolates by animal sources, and more accurate in classifying query isolates to the correct animal sources . Our data suggest that, by taking a snapshot of the system-wide expression of bacterial cells and circumventing peptide identification, the novel method generated fingerprints that not only reflected the adaptation of E. coli to different animal hosts more precisely than PCR-DNA fingerprinting but also constituted a fuller representation of the bacterial cells' proteomes than traditional shotgun proteomics.