Theory and evidence suggest that some selective pressures are more common on islands than in adjacent mainland habitats, leading evolution to follow predictable trends. The existence of predictable evolutionary trends has nonetheless been difficult to demonstrate, mainly because of the challenge of separating in situ evolution from sorting processes derived from colonization events. Here we use brain size measurements of >1900 avian species to reveal the existence of one such trend: increased brain size in island dwellers. Based on sister-taxa comparisons and phylogenetic ancestral trait estimations, we show that species living on islands have relatively larger brains than their mainland relatives and that these differences mainly reflect in situ evolution rather than varying colonization success. Our findings reinforce the view that in some instances evolution may be predictable, and yield insight into why some animals evolve larger brains despite substantial energetic and developmental costs.