Benthic foraminiferal thanatocoenoses from the Cap-Ferret Canyon area were studied in the >150-µm fraction of 4–5 cm deep sediment levels, at 13 stations. The shallowest station (151 m depth) is located at the shelf break, close to the canyon head. All other stations are located along two bathymetric transects: seven stations along the canyon axis between 300 and 3000 m depth, and five stations from 300 m to 2000 m depth along the southern flank of the canyon. The comparison between the live (Rose-Bengal-stained) and dead assemblages shows that biological (i.e. population dynamic) and taphonomic processes (i.e. test destruction, transport) generate important discrepancies between live and dead assemblages. An important question is, to what degree post-mortem transport and redeposition of foraminiferal tests contribute to the difference between living and dead assemblages? The composition of the thanatocoenoses (<1% of neritic species) indicates that there is no substantial transport of shells >150 µm from the inner continental shelf to the Cap-Ferret Canyon axis. However, transport of tests from outer shelf or upper canyon axis towards deeper sites occurs, as indicated by an increase of diversity indices of the dead fauna along the canyon axis. Moreover, some species (e.g., Cassidulina carinata) are observed in the living fauna restricted to the shallow sites, but occur in important amounts in the dead fauna at deeper stations, suggesting that these taxa have been transported from upper canyon stations toward deeper sites. Since Cap-Ferret Canyon is inactive in terms of massive sediment transport (i.e. gravity events), downslope transport of foraminiferal tests probably takes place in nepheloid layers. Downslope transports of foraminiferal tests may create important biases for the utilisation of paleoceanographic proxies using the assemblage characteristics and/or the geochemical composition of selected species. However, the study of dead assemblages along a canyon axis can give important clues about the sedimentary dynamics, especially an idea of the quantity of reworked foraminifera transported through the canyon axis. Along the adjacent flank, such evidence of downslope transport was not observed. Despite taphonomic (test destruction) and biological processes (population dynamics), the live fauna closely resembles the distribution and the composition of the dead assemblage. Therefore, ecological information obtained from the fossil flank samples can be more reliably used to reconstruct paleoenvironmental conditions.Read less <