Sensorial characterisation is extremely valuable for products benefiting from "AOC" status. However, the methods generally used for officially guaranteed food products are poorly adapted to "AOC" products. This is because products belonging to the same "AOC" are inevitably highly variable. The characterisation of these products cannot therefore be limited to a single sensorial profile but rather requires the construction of several different profiles. This does not in anyway question the validity of the "AOC" designation. Paradoxically, the small volume produced of many AOC products prevents the construction of these sensory profiles for economic reasons. In this article, the authors describe how the current empirical approach works, using the example of cheese products. This approach is based on a simple sensorial method involving the professional sector. The authors also consider how this approach may be validated and improved.