The response of species diversity to dispersal capability is inherently scale-dependent: increasing dispersal capability is expected to increase diversity at the local scale, while decreasing diversity at the metacommunity scale. However, these expectations are based on model formulations that neglect dispersal limitation and species segregation at the local scale. We developed a unifying framework of dispersal–diversity relationships and tested the generality of these expectations. For this purpose we used a spatially-explicit neutral model with various combinations of survey area (local scale) and landscape size (metacommunity scale). Simulations were conducted using landscapes of finite and of conceptually infinite size. We analyzed the scale-dependence of dispersal-diversity relationships for exponentially-bounded versus fat-tailed dispersal kernels, several levels of speciation rate and contrasting assumptions on recruitment at short dispersal distances. We found that the ratio of survey area to landscape size is a major determinant of dispersal–diversity relationships. With increasing survey-to-landscape area ratio the dispersal–diversity relationship switches from monotonically increasing through a U-shaped pattern (with a local minimum) to a monotonically decreasing pattern. Therefore, we provide a continuous set of dispersal–diversity relationships, which contains the response shapes reported previously as extreme cases. We suggest the mean dispersal distance with the minimum of species diversity (minimizing dispersal distance) for a certain scenario as a key characteristic of dispersal–diversity relationships. We show that not only increasing mean dispersal distances, but also increasing variances of dispersal can enhance diversity at the local scale, given a diverse species pool at the metacommunity scale. In conclusion, the response of diversity to variations of dispersal capability at spatial scales of interest, e.g. conservation areas, can differ more widely than expected previously. Therefore, land use and conservation activities, which manipulate dispersal capability, need to consider the landscape context and potential species pools carefully.