Abstract Differences between marine soft-sediment assemblages at disturbed (two waste dumps, a sewage outfall and a wharf) and control locations were found at Casey Station, Antarctica. These differences were significant against considerable background spatial variability. Core samples were collected by divers using a hierarchical, spatially nested sampling design incorporating four scales: (i) locations (thousands of metres apart); (ii) sites (hundreds of metres apart); (iii) plots (tens of metres apart); and (iv) among replicates within plots (approximately 1 metre apart). Control locations had greater species richness and diversity than disturbed locations and there were many taxa found at control locations that were not recorded at disturbed locations. Assemblages at disturbed locations were less variable than those at control locations. In contrast, populations of some dominant species were more variable at disturbed locations than at control locations. Significant variation in populations of individual taxa was also found at all scales and although greatest at the level of location, variation was also large at the smallest scale, between replicate cores, indicating significant small-scale patchiness in populations of taxa. Patterns of assemblage structure were similar at fine (77 taxa) and medium (aggregated to 33 taxa) levels of taxonomic resolution, but changed at coarse levels of resolution (nine phyla). Soft-sediment assemblages at Casey Station are markedly different from those reported from other areas of Antarctica. Assemblages at Casey are almost completely dominated by crustaceans (up to 99% total abundance) and polychaetes are rare or in very low abundances in most areas investigated in the region. This is also the first demonstration that small Antarctic research stations (population 20–50 people) may cause impacts that are detectable in the adjacent marine environment.