The traditional microbiological approach to isolating and identifying the bacteria associated with food fermentations has permitted us to gain some understanding of the processes involved.  However, it has become increasingly clear over recent years that this understanding is incomplete.  The advent of PCR-based technology has indicated that many products contain not only microbes that are viable but not culturable, but also previously unidentified species. Furthermore, electrophoretic separation of DNA fragments indistinguishable in size by conventional agarose gel electrophoresis has been made possible by combining a denaturing gradient with the separation matrix. This can be achieved using either a temperature gradient (TGGE) or a chemical gradient (DGGE). Gels of PCR products derived from variable regions of 16S RNA gene sequences show the presence of a range of fragments with differing mobilities, each of which may represent a single bacterial species or sub-species. Gels from different samples can be compared and used to show the development of populations within a food substrate during fermentation. Bands can then be excised from a gel and sequenced thereby allowing the population to be tentatively identified. The challenge will then be to use these data to predict how the recovery media can be adapted to permit the cultivation of the key microbes so that their characteristics can be studied in vivo and the fermentation processes optimised.