During the Palaeolithic, the cultural territory is traditionally represented for each prehistoric group by the area containing all sites where evidences of this material culture occur. Inside this large cultural territory we could attempt to identify provisioning territories by studies focused on one single site or by regional synthesis showing the relations between several places. However, some studies highlight that provisioning territories and cultural territories do not always fit. How to interpret such results? Here we present a cross-study of four French sites from both Middle and Upper Palaeolithic in order to compare the classical model and the outsider results. We will first synthesis the results of Combe-Grenal (Lot; Middle Palaeolithic) and l'abri Pataud (Dordogne; Upper Palaeolithic) that fit with the classical model and show clearly all the well-known differences regarding provisioning territories between these two phases. Then we will detail two other cases that show significant differences between provisioning and cultural territories: Baume-Vallée (Haute-Loire; Middle Palaeolithic) and le Pigeonnier (Gironde; Upper Palaeolithic). Finally, we will discuss the interpretation of these results on various levels questioning:-our own concepts: what is a cultural territory and how do we define the share of knowledge through space?-on what data are we basing the comparisons to support the cultural territory expansion?-and finally the palethnological implications of these outsider cases for both Middle and Upper Palaeolithic: could we precise the management of raw materials between direct (nomadism) or indirect (trade) provisioning by the study of resources and/or groups mobility?