Despite edges being common features of many natural habitats, there is little general understanding of the ways assemblages respond to them. Every edge between two contrasting habitats has characteristics governed by the composition of adjoining habitats and/or by the nature of any transitions between them. To develop better explanatory theory, we examined the extent to which edges act independently of the composition of the surrounding landscape and to which transitions between different types of habitats affect assemblages. Using experimental landscapes, we measured the responses of assemblages of marine molluscs colonising different experimental landscapes constructed with different compositions (i.e. different types of habitats within the landscape) and different types of transitions between habitats (i.e. sharp vs gradual). Edge effects (i.e. proximity to the edge of the landscape) were independent of the internal composition of experimental landscape; fewer species were found near the edges of landscapes. These reductions may be explained by differences in differential larval settlement between edges and interiors of experimental landscapes. We also found that the sharpness of transitions influenced the magnitude of interactions in the different types of habitats in experimental landscapes, most probably due to the increased number of species in areas of transition between two habitats. Our experiments allowed the effects of composition and transitions between habitats to be disentangled from those of proximity to edges of landscapes. Understanding and making predictions about the responses by species to edges depends on understanding not only the nature of transitions across boundaries, but also the landscape in which the edges are embedded.