We present in this paper a Blind Source Separation (BSS) application to the separation and identification of microorganism Raman spectra. In the study of microorganisms by Raman spectroscopy, the major difficulty is to dissociate the Raman spectrum (RS) of the bacteria from that of the solid culture medium, on which they grow. This is classically achieved by measuring the pure solid culture medium RS and subtracting its exact value from the bacteria and medium RS. The major advantage of the BSS techniques is the extraction of the bacterial RS without prior measurement from the pure solid culture medium, which permits to be independent of medium variations.