Freshwater ecosystems have been fragmented by the construction of large numbers of dams. In addition to disruption of ecological continuity and physical disturbance downstream, accumulation of large amounts of sediment within run-of-river reservoirs constitutes a latent ecotoxic risk to aquatic communities. To date, run-of-river reservoirs and ecotoxic risks associated with contaminated sediment to the biodiversity and functioning of such systems are little studied. Therefore, the main objective of this study was to describe macroinvertebrate assemblages, and the functioning of these systems, and to propose indicators of sediment contamination to integrate in in-situ risk assessment methodology. To identify specific assemblages of run-of-river reservoirs, we first compared macroinvertebrate assemblages and their biotrait profiles (i.e. from a database of biological and ecological traits) in reservoirs (n = 6) and associated river sites (upstream and downstream of dams). Then, we compared responses of assemblages and biotrait profiles to sediment contamination of the banks and channels of reservoirs to select the most useful spatial scale to identify sediment contamination. Nineteen indicator taxa were observed to be specifically associated with channel habitats of reservoirs. Among these, the abundance of three taxa (Tanypodinae (Diptera), Ephemerella (Ephemeroptera) and Atherix (Diptera)) revealed the effect of metal sediment contamination. "Between-reservoirs" differences in their biotrait profile were found along the contamination gradient, with a shift of communities' composition and functionality, and an increase in functional similarity. Many traits (response traits), for example "maximum size", "transverse distribution", "substrate preferences", "saprobity", "temperature", "resistance forms", and "locomotion", were specifically linked to contamination of sediments by metals. This study indicates how sediment contamination can change the structural and functional composition of run-of-river reservoir assemblages. Indicator taxa and response traits identified in this study could improve current risk assessment methodology and potentially enable prediction of the risks of contaminated sediments stored in reservoirs in downstream ecosystems.