Benthic macrofauna communities were sampled with their 'undisturbed' sediments from 2 soft-bottom sites on the Swedish west coast. At one of the sites the top sediment had a positive redox potential (Expt 1) and at the other the sediment was comparatively more reduced close to the surface (Expt 2). The benthic communities at these 2 sites were each exposed to 3 treatments of oxygen concentrations: (1) normoxia (>8.0 mg l−1, >90% oxygen saturation), (2) moderate hypoxia (∼1.0 mg l−1, 12% oxygen saturation) and (3) severe hypoxia (∼0.5 mg l−1, 6% oxygen saturation). Redox potential profiles in the sediments showed a significant increase with time under normoxic conditions, whereas they showed an overall decline with time in moderate and severe hypoxia. The number of individuals and species at the 2 field sites did not differ significantly from corresponding numbers in the experimental normoxia (laboratory control) at the end of those experiments. In Expt 1 survival of the number of individuals was reduced in severe hypoxia compared to in moderate hypoxia and normoxia. The number of species was significantly reduced between all treatments. In Expt 2 community abundance and the number of species were significantly reduced in moderate and severe hypoxia compared to normoxia. Behaviour of dominant species is described: in general they left their position in the sediment (for Pectinaria koreni with their tubes) and occurred at the sediment surface in the moderate and severe hypoxic treatments. Comparison of the results of Expt 1 with those of Expt 2 suggests that the comparatively more reduced sediment in Expt 2 resulted in a greater negative response, implying the significance of sediment conditions for benthic macrofauna response to hypoxia.