The earliest known complete bats, from the Eocene (49–53 Mya), were already capable of flapping flight and echolocation. In the absence of direct fossil evidence there have been many speculative scenarios advanced to explain the evolution of these behaviours and their distributions in extant bats. Theories assuming chiropteran monophyly have generally presumed the ancestral pre-bat was nocturnal, arboreal and insectivorous. Following this assumption hypotheses can be divided into the echolocation first, flight first and tandem development hypotheses, all of which assume that flight evolved only once in the lineage. In contrast, the chiropteran diphyly hypothesis suggests that flight evolved twice. Evidence supporting and refuting the different hypotheses are reviewed. It is concluded that there are significant problems attached to all the current models. A novel hypothesis is advanced, which starts from the assumption that bats are monophyletic and the ancestral pre-bat was arboreal, but diurnal and frugivorous. After the evolution of flight it is suggested that these animals were driven into the nocturnal niche by the evolution of raptorial birds, and different groups evolved either specialised nocturnal vision (megachiropterans) or echolocation (microchiropterans). A block on sensory modality transfer has retained this distribution of perceptual capabilities ever since, despite some Megachiroptera evolving rudimentary echolocation, and the dietary convergence of some Microchiroptera with the Megachiroptera. The new hypothesis overcomes many of the problems identified in previous treatments.