Although macroecology arose from geographical ecology, it has diverted from a geographical perspective. At present, most macroecological studies use a statistical approach that adopts an ‘individual species focus’ and relies on comparisons between species to test for broad-scale ecological patterns. Sometimes, space is included as part of the analysis, but almost always in a single dimension. In both situations, observed relationships are depicted using bivariate scatter-plots. We argue that current macroecological approaches may interfere with our perception of patterns and have important implications for their biological interpretation. We use the literature concerned with spatial variation in the range sizes of species (Rapoport's rule) to illustrate our point of view. Given the current lack of maps actually showing the patterns we are trying to explain, we contend that macroecology could benefit greatly by returning to its geographical roots, at least when data contain spatial structure.