The bony labyrinth consists of three parts (the two vestibular sacs, the three semicircular canals and the cochlea) and houses two functional systems. The vestibular system provides one way of motion detection in a three-dimensional space. The cochlea is specialized for sound detection. The close anatomical relationship between the bony labyrinth and the corresponding receptor endorgans provides an opportunity to study osteological specimens (including fossils). The investigation of the three-dimensional (3D) anatomical variation of the bony labyrinth in extant species represents a prerequisite for the interpretation of their fossil closest relatives. This prerequisite has not been completely fulfilled yet due to (i) the 3D geometry complexity of the labyrinth ; (ii) the difficulty to acquire high resolution data ; (iii) the few proposed expert-independent comparative methods. Here, we use 3D geometrical models of 40 bony labyrinths reconstructed from micro-CT scans of extant humans, chimpanzees, bonobos and baboons. We use recently developed automated computational tools allowing to process 3D free-form surfaces, and more specifically to assess the mean anatomy within a sample, the pattern of variability around this mean, and to compare samples. Our results allow us to identify the most common features and most striking differences within and between species. The use of such automated, 3D and objective techniques, coupled with standard linear, surface or volume measurements, may allow to gain further insight into the co-evolution of the two functional systems housed in the inner ear of the extant and extinct higher primates.