Determining both the genetic causes and the functional consequences of morphological variation is critical to understand how organisms adapt to their local environment. Like a number of other rodent taxa, deer mice (Peromyscus maniculatus) that live in forested habitat have evolved longer tails than prairie mice of the same species. This difference evolved separately at least twice, in eastern and western North America. Differences in tail length are proposed to improve performance during arboreal locomotion; testing such functional hypotheses is critical to understand adaptation. Here, using two replicate forest-prairie subspecies pairs, we tested both the genetic and developmental mechanisms that underlie the longer tails of forest mice, and the functional consequences of the tail length difference for balance. We found that forest deer mice consistently perform better in a simple assay of arboreal locomotion, even when reared in the lab and naïve to climbing. In both eastern and western subspecies, genetically and developmentally independent changes in vertebra length and number contribute to the longer tails of forest mice. Despite these highly parallel phenotypes, we found that the underlying causative alleles are likely at least partly distinct in eastern and western subspecies. Finally, we used hybrid, laboratory-reared populations alongside simple analytical models to test the functional significance of tail length; using this approach, we found that the inter-population and inter-individual differences in tail length may contribute to performance. By addressing both the genetic causes and functional consequences of parallel evolution within species, this work provides insight into the mechanisms of local adaptation.