Stomach content analyses are commonly used to study both fish feeding behaviour and trophic conditions. However, the interpretation of such data depends on fish foraging behaviour for a given environment and how representative the stomach contents are to the prey distribution. Tuna feeding behaviour was studied within the context of a research programme conducted in French Polynesia. Tuna prey distribution was characterised using acoustic measurements and pelagic trawls; thereafter, this distribution was compared with the stomach contents of tuna caught using an instrumented longline. Acoustic, pelagic trawling and stomach content analyses give complementary elements to describe the pelagic trophic habitat and to better understand tuna-prey relationships. The classic concept of a reduced food availability for tunas in the tropical pelagic environment seems relative. Tunas able to dive enough during daytime to exploit the migrant micronektonic species secure a source of regular food. This is particularly true of bigeye tuna (Thunnus obesus), which have ecophysiological capacities for this purpose. The behaviour of albacore tuna (T. alalunga), which dive > 400 m in depth, remains less clear, as little is known about their vertical behaviour. Lastly, yellowfin tuna (T. albacares), which are distributed in more superficial waters, can better exploit the biomass of juvenile fish and crustaceans exported from the reefs. Analysis of the stomach fullness of tuna caught by longline, a passive gear, generally showed an empty state. This result suggests that most tuna foraging on large prey aggregations present in the study area are quickly satiated and escape longline capture and sampling. A consequence is that studies of tuna feeding behaviour based on longlining may be biased, particularly when large aggregations of prey are present such as in convergence zones. Another potential consequence is that longline tuna catch tares could differ according to prey richness. Longline tuna catch rates may sometimes reflect the relative abundance of prey rather than relative tuna abundance. Electronic supplementary material to this paper can be obtained by using the Springer LINK server located at http://dx.doi.org/10.1007/s00227-001-0776-3.