Diprotodontia represents the largest and ecologically most distinct order of marsupials occurring in Australasian being highly divers in size, locomotion, habitat preferences, feeding, and activity pattern. The spatial orientation in the habitat and therefore the three-dimensional space is detected by the vestibular system of the inner ear, more precisely by the three semicircular canals. In this study, we investigated the bony labyrinth of diprotodontian and selected non-diprotodontian marsupial mammals of almost all genera with noninvasive micro-CT scanning and 3D-reconstructions. In principal component analyses, the subterranean taxon can be separated from gliding and saltatorial taxa, whereas arboreal species can be separated from saltatorial specimens. The highest PCA loadings of this functional distinction are clearly found in the diameter of the semicircular canals, whereas the overall shape (height, width, length) of the semicircular canals is less important. Additionally, the investigated arboreal and fossorial species of South America are nested in the morphospace of the Australasian taxa. Even if a phylogenetic signal in the anatomy of the bony labyrinth cannot be excluded entirely, the main functional morphological signal of the vestibular system is found in the diameter of the semicircular canals. With the large dataset of extant marsupial mammals analysed here, the locomotion mode of extinct taxa can be inferred in future studies independent of any evidence of postcranial material.