Habitat clustering results from processes of habitat loss and fragmentation, which operate at different resolutions and with different intensities, e.g. forest clear-cutting or thinning. Individual movements also vary at different spatial scales according to landscape structure and species dispersal strategies. Disentangling the relative impact of habitat loss and fragmentation on the long-term survival of species requires understanding how clustering at one resolution interacts with the amount of habitat, dispersal distance and clustering at other resolutions, to affect dispersal success. We addressed this problem by quantifying the magnitude of these interactions and how they were affected by the intensity of habitat removal. Individual-based simulations were conducted on artificial fractal landscapes where the intensity of habitat removal and the amount of clustering were varied independently at two nested resolutions, while the total amount of habitat in the landscape was controlled for. We show that the way the amount of habitat, the dispersal distance and the amount of clustering affect dispersal success depends on the resolution at which habitat clustering occurs, the intensity at which habitat is removed, and the strength of habitat selection. Our findings highlight: (a) the importance of explicitly considering scale-dependent biological responses to landscape change; and (b) the need to identify the appropriate scale at which to manage fragmentation, thus avoiding mismatches between the scale of ecological processes and the scale of management.