Surface-enhanced Raman spectroscopy (SERS) has been suggested as a powerful tool to identify bacteria, drawing from its high fingerprint (vibrational) information content, its extreme sensitivity (down to the single molecule level) and its obliviousness to the aqueous environment intrinsic to biological systems. We review here in a comparative manner the various studies that attempted to utilize SERS for this important goal in light of the work carried out by our own group over the past 10 years or so. We show that SERS has an additional major advantage, namely, it introduces a new dimension of selectivity, which, on the one hand, makes it even more suitable as an analytical tool, but on the other hand, it requires gaining control of the precise manner in which the SERS-active metal centers are produced and brought into contact with the micro-organism. Our emphasis in this review is on understanding the spectra in terms of the nature of the SERS-active centers and their placement within the bacterium. On the interpretation and assignment of the spectra, we constantly keep in mind the final goal of bacteria identification.