Evolutionary thinking can inform the choice and assessment of model species in neuroscience, particularly when such models are intended to generate knowledge that will translate to humans. Avoiding errors that arise from oversimplified notions of phylogeny or genotype-phenotype mapping is one contribution; evolutionary biology also offers positive guidance. The challenge of finding adequate non-human models for translational research is particularly acute in neuroscience: neurobiological and behavioral phenotypes are complex and plastic, and many traits important in humans are absent, radically different, or difficult to assess in other species. Evolutionary perspectives help to articulate and address these challenges. Darwin’s description of “descent with modification” points to two aspects of evolution that can help us assess the matching between a prospective model species and its intended target. One is trees that represent the structure of phylogenetic relationships; the other is phenotypic traits, i.e. the unique characteristics of each species’ evolved biology and natural history. Mapping traits onto a phylogeny is the first step toward analyzing the source of similarities between a target and a potential model. Whether similar traits arise from shared ancestry or from adaptive convergence has important implications for what kinds of inferences can be justified, and for the likely translatability of findings. Evolution offers both a rich source of possible models, and guidance for choosing the best ones for a given purpose. Considering model choice from an evolutionary angle not only helps to answer the question “What species might be a good model for studying x?” but also suggests additional questions we should be asking to assess the utility of both potential and current models. Recognizing the diverse ways model organisms can function expands our search image as we seek species to study that can both extend general knowledge, and generate translatable insights relevant to human neurobiology and disease.