An important ecological issue is developing an understanding of how patterns and processes vary with scale. We designed a field experiment to test how differences in the aerial extent of disturbance affected macrofaunal recolonization on a sandflat. Three different plot sizes (0.203 m 2 , 0.81 m 2 , and 3.24 m 2 ) were defaunated, and samples were collected to assess recovery over a 9-mo period. As the sandflat used for the experiment was prone to disturbance by wind-driven waves, we also measured changes in sediment bed height (an indicator of sediment stability) over the course of the experiment. Most common species revealed significant relationships between density and disturbance plot size. Scale-dependent recovery was also demonstrated by differences in species assemblage structure over the course of the experiment. Relative rates of colonization varied by 50% between large and small experimental plots. However, these differences were not related to specific species, particular functional groups, or potential modes of colonization. The results revealed an unusually slow rate of faunal recovery following defaunation. Both increasing numbers of colonists and density changes in ambient sediments made an important contribution to recovery. The relationship found between changes in sediment bed height and wind velocity indicated that wind-driven wave disturbance was an important factor influencing sediment instability. Sediment instability was higher in all experimental plots than in the ambient sediments, due to the initial removal of a dense spionid polychaete tube mat characteristically found at the study site. Sediment instability also increased with increasing plot size. Thus in this dynamic sandflat habitat, faunal emigration from recovering disturbed patches of sediment may significantly slow rates of recolonization. These results demonstrate that incorporating patch size, emigration, recovery time, and interactions between hydrodynamic conditions and habitat stability (particularly where colonists influence sediment stability) are crucial to generating a general understanding of recovery processes in soft-sediment habitats. While our results demonstrate the need for caution in scaling-up from small-scale studies, they do indicate that larger scale disturbances that destroy organisms with a role in maintaining habitat stability are likely to result in very slow recovery dynamics, particularly in wave-disturbed soft-sediment habitats.