Concerning the isolation of volatile compounds, many instrumental analyses of volatile flavours in food have been carried out without considering changes that may occur during eating as temperature increase, salivation and mastication. The development by Lee (1986) of a model system able to monitor the headspace continuously by using a mass spectrometer coupled with a dynamic headspace system was followed by many other attempts to reproduce the oral processing of food. Roberts and Acree (1996) presented a model in which food was macerated and hydrated under defined conditions and the volatiles released into the gas phase were analysed. van Ruth, Roozen, and Cozijnsen (1994) developed a model system in which mouth conditions are simulated taking into account the volume of the mouth, temperature, salivation and mastication. The performance of this model regarding the extraction of volatile compounds was compared to a dynamic headspace system, a purge and trap system and a direct sampling in the mouth of assessors owing to a Tenax trap (van Ruth & Roozen, 2000). The performance of extraction was checked on the release of aroma compounds of rehydrated French beans. It was demonstrated that aroma release in the artificial mouth system did not differ significantly from release in the mouth of the 12 assessors used for direct sampling. Both the release in the artificial mouth and the release in the mouths of assessors differed significantly from the ones obtained with the other methods.