To understand both the effect of a physical modification of a river course by dams and of a metallic contamination of sediment within dams, macrobenthic communities of an alpine stream were studied. Macroinvertebrates were sampled on five sites along the same river course: an upstream reference site, a reference storage, a site downstream this first storage, a metal-contaminated storage and a downstream site. Sampling protocol allowed us 1) to assess both faunistic and functional modifications from the habitat to the station scale and 2) to define the respective impacts upon banks and channel communities. Our results showed that bank samples are more efficient than channel ones to evidence an upstream downstream gradient, with the contaminated dam clearly outside the general pattern, considering both faunistic and functional information. Considering the habitat scale, we showed that both erosion and sedimentation habitats allowed to evidence the existence of a contaminated site based on the faunistic composition. More surprisingly, it appeared that assemblages found in our storages did not showed particular and specific characteristics compared to stream assemblages.