Species are differentially affected by habitat fragmentation as a consequence of differences in mobility, area requirements, use of the matrix, and responses to edges. A quantitative understanding of these differences is essential not only for conservation biology but also for basic ecological theory. Here, we examine density responses by butterflies to patch size and use a quantitative theory on the scaling of population density with patch size to interpret results. Theory suggests that the density distribution of mobile species along a patch size gradient should depend on the scaling of net migration rates, whereas the density distribution of less mobile species should depend more on local growth. Using data from 11 localities in three European countries, we calculated the slope in the relationship between patch size and population density. These slopes were evaluated in relation to butterfly traits and matrix composition. As estimates of butterfly mobility we used both wing span and expert mobility rankings. The slope of the density–area relationship changed as predicted with wing span and the association of species to grasslands. Large and highly mobile species had a negative slope, similarly for grassland specialists and generalist species, and the slope matched quantitative predictions based on the scaling of net migration rates. Small and less mobile grassland specialists had a slope that was less negative than the slope of large and mobile grassland specialists, whereas the slope did not change with size for generalist species. These analyses suggest that the variability in response among butterfly species to patch size could be explained by accounting for body size/mobility and habitat associations among species. A caveat is that edge effects are not explicitly included in the model analysis, and future research should aim to combine area and edge effects in a common theoretical framework.