The origin and early evolution of snakes has long been studied, but little research has focused on soft-tissue organs such as the brain. I report data from dissections and 3D reconstructions of the endocasts of diverse species, including the Cretaceous stem snake Dinilysia patagonica in order to provide a comparative evolutionary framework for the snake brain. Snakes are a special case among reptiles because the braincase almost entirely encloses the whole brain, so endocasts provide realistic representations of brain size and shape. Diversity of brain gross anatomy among snakes is remarkable, encompassing two major cerebrotypes occurring in surface-dwelling and burrowing species. The repeated acquisition of the burrowing cerebrotype in different and phylogenetically distant snake clades suggests that brain gross anatomy is surprisingly evolutionary labile in snakes. Brain gross anatomy and other features such as body size and the absence of any unequivocal osteological feature related to burrowing is interpreted as evidence that D. patagonica was surface-dwelling, and that at least some of the early history of snakes occurred above ground.