Uncovering relationships between neuroanatomy, behavior, and evolution are important for understanding the factors that control brain function. Voluntary exercise is one key behavior that both affects, and may be affected by, neuroanatomical variation. Moreover, recent studies suggest an important role for physical activity in brain evolution. We used a unique and ongoing artificial selection model in which mice are bred for high voluntary wheel-running behavior, yielding four replicate lines of high runner (HR) mice that run ∼3-fold more revolutions per day than four replicate nonselected control (C) lines. Previous studies reported that, with body mass as a covariate, HR mice had heavier whole brains, non-cerebellar brains, and larger midbrains than C mice. We sampled mice from generation 66 and used high-resolution microscopy to test the hypothesis that HR mice have greater volumes and/or cell densities in nine key regions from either the midbrain or limbic system. In addition, half of the mice were given 10 weeks of wheel access from weaning, and we predicted that chronic exercise would increase the volumes of the examined brain regions via phenotypic plasticity. We replicated findings that both selective breeding and wheel access increased total brain mass, with no significant interaction between the two factors. In HR compared to C mice, adjusting for body mass, both the red nucleus (RN) of the midbrain and the hippocampus (HPC) were significantly larger, and the whole midbrain tended to be larger, with no effect of wheel access nor any interactions. Linetype and wheel access had an interactive effect on the volume of the periaqueductal gray (PAG), such that wheel access increased PAG volume in C mice but decreased volume in HR mice. Neither linetype nor wheel access affected volumes of the substantia nigra, ventral tegmental area, nucleus accumbens, ventral pallidum (VP), or basolateral amygdala. We found no main effect of either linetype or wheel access on neuronal densities (numbers of cells per unit area) for any of the regions examined. Taken together, our results suggest that the increased exercise phenotype of HR mice is related to increased RN and hippocampal volumes, but that chronic exercise alone does not produce such phenotypes.