The use of in vitro techniques is rapidly expanding because of an increasing need for routine and reproducible methods to obtain bioavailability data in addition to chemical ones. Furthermore in vivo methods are very expensive, an invasive technique for animals and may give non homogeneous results. In vitro systems could be divided in simulation systems (e.g. Rusitec, Czerkawski and Brekenridge, 1977) that attempt to closely reproduce the animal rumen ecosystem and are characterised by a higher level of complexity; and techniques that are mainly targeted to obtain biavailability data, without trying to mimic rumen metabolism (e.g. enzyme tests, Antoniewicz and Kosmala, 1997). As the fate of feedstufis in the rumen is mostly regulated by the micro-organisms, a rumen simulation system should primarily be characterised by a representative rumen microflora. The aim was to evaluate the possibility of maintaining a normal rumen microflora in an in vitro simulation system.