Bone retouchers are the most ancient osseous tools of the European Palaeolithic. They are very frequent on Mousterian sites and persist throughout the Upper Palaeolithic. Several functional analyses have been devoted to them and the most commonly accepted hypothesis proposed is their use for lithic tool retouching by percussion. However, this hypothesis is based mainly on studies of Mousterian pieces while few functional analyses have been undertaken on Upper Palaeolithic bone retouchers. The persistent presence of these “ ad hoc” tools makes their study important. This article presents the results of a techno-functional analysis devoted to the Early Aurignacian bone retouchers from Abri Castanet at Sergeac (Dordogne), northern and southern sectors (layer A and US 131 and below), Grotte des Hyènes at Brassempouy (Landes, complex 2) and Gatzarria at Suhare (Pyrénées-Atlantiques, layer cbcf). These unworked bone tools, including many specimens identified a posteriori among faunal remains, represent a significant part of the studied assemblages’ osseous equipment. There are two categories of retouchers : those on raw bone fragments and those on technical objects. The former, most numerous, are mainly long bone shaft fragments recovered after food operations and used just as they were, without modification. The latter correspond to the use of finished objects like smoothers and awls as well as manufacturing by-products. They bear diagnostic traces of utilisation in the form of linear, deep, short depressions, v-shaped in cross section. These depressions are similar to those that have been described in previous work involving microscopic observation and are compatible with use for lithic tool retouching by percussion. However, unlike Middle Palaeolithic retouchers, where the depressions are perpendicular to the main axis of the piece, Early Aurignacian depressions are often vertical (in combination or not with horizontal ones). This is a characteristic feature of Upper Palaeolithic retouchers related to their positions during use. The vertical depressions imply that the main axis of the retoucher is oriented parallel to the lithic edge during use. Quite obviously, this change in the functioning of retouchers must be linked to changes in flint-knapping processes. To understand this relationship, we explored the effects of changing the orientation of retouchers. One notable difference is their curvature in the transverse and longitudinal axes. The latter is clearly more pronounced than the former, meaning that the retouchers offer a greater overall surface area when used in a parallel orientation. We hypothesized that this orientation was preferred to increase the active surface of retouchers for the retouching of blade products. Indeed, during the Early Aurignacian, the manufacture of blade tools, scrapers in particular, often involved retouching very convex and narrow ends. Orienting the retoucher parallel to the lithic edge would have been a technical solution to compensate for the decrease in the active surface presented by the strong convexity of blade ends. However, within the Early Aurignacian lithic industry, blade ends are not the only items to show strong convexity : the fronts of carinated and nosed scrapers also display them. The use of retouchers in bladelet knapping has not so far been considered, but constitutes a second hypothesis that could explain the particular orientation of Early Aurignacian retouchers during use. T hese two assumptions have been tested by experimental replication. Results show that bone retouchers are effective tools for both blade retouching and bladelet knapping, provided that the weight of the retoucher is adjusted to suit the type of lithic product. In addition, lithic and bone tools produced experimentally fall within the range of variation observed in the archaeological assemblages studied. Finally, although this has not always been necessary, experimentation confirms the usefulness of parallel retoucher orientation to adapt the active surface areas to strong lithic convexity. To date, strong convexity of lithic products is the main factor likely to explain the change in retoucher orientation that occurred at least as early as the Early Aurignacian. Thus, retouchers of this period attest to the duration of an ancient technical process, involving the use of bone for stone knapping. Specimens bearing vertical depressions thus illustrate the adaptation of the retouchers to “ new” lithic technologies. The important role played by retouchers in retouching Mousterian tools and their number within the assemblages studied leave little doubt as to their use in retouching Upper Palaeolithic blade products. However, it is still difficult to determine whether they were in fact used in bladelet knapping. The traces of utilisation on experimental retouchers are quite similar to those on specimens used in blade retouching. An expanded experimental program and analysis of the bone tools with high magnification should facilitate the assessment of morphometric changes in use-wear traces based on activity. In addition, a comparison of technical and functional characteristics of archaeological and experimental lithic products should help clarify the array of hammers used during the Aurignacian.