The peak echolocation frequency of insectivorous bats generally declines as body size increases. However, there are notable exceptions to this rule, with some species, such as Rhinolophus clivosus, having a higher than expected peak frequency for their body size. Such deviations from allometry may be associated with partitioning of foraging habitat (the foraging habitat hypothesis) or insect prey (the prey detection hypothesis). Alternatively, the deviations may be associated with the partitioning of sonar frequency bands to allow effective communication in a social context (the acoustic communication hypothesis). We tested the predictions of these hypotheses through comparisons at the family, clade and species level, using species of rhinolophids in general and R. clivosus, a species with a wide distribution, as a specific test case. We compared the wing parameters, echolocation frequency and ecology of R. clivosus to those of the sympatric R. capensis. Rhinolophus clivosus has a much higher echolocation frequency than predicted from its wing loading or body mass. Furthermore, contrary to the predictions of the foraging habitat hypothesis, we found no difference in foraging habitat between R. clivosus and R. capensis. The size range of insect prey taken by the two species also overlapped almost completely, contrary to the prey detection hypothesis. On the other hand, the variation of echolocation frequencies around the allometric relationship for rhinolophids was smaller than that for Myotis spp., supporting the prediction of the acoustic communication hypothesis. We thus propose that the relatively high peak frequency of R. clivosus is the result of partitioning of sonar frequency bands to minimize the ambiguity of echolocation calls during social interactions.