Sagittal otolith shape analysis is one of the most widespread techniques worldwide to discriminate fish stock units, as this proxy integrates both environmental and genetic factors. All previous otolith shape studies have been carried out using two-dimensional (2D) images, a partial representation of the whole shape of the otolith. However, prior to the identification of stock unit boundaries, the influence of other potential drivers controlling the otolith shape must be analysed to limit their bias. In this study, the presence of asymmetry in otolith shape depending on the inner ear side (i.e., left versus right inner ears) was tested by comparing the approaches of 2D and three-dimensional (3D) sagittal otolith shape analyses. Eighty-two red mullet adults (Mullus barbatus) from three locations in the eastern part of the Mediterranean Sea were studied. Fourier harmonic descriptors computed from 2D outlines and spherical harmonic descriptors computed from 3D meshes were used to evaluate otolith shape variation. The results of a multivariate mixed-effects model from 2D images showed that there was no asymmetry effect of inner ear side on the otolith shape in any location. There was, however, a significant geographical effect for the 2D otolith shape between the Adriatic Sea and the Levantine Sea. In contrast, 3D information showed that both side effects and geographical differences were significant. This is the first study comparing 2D and 3D data showing different results on the same sample of red mullet. These results demonstrate the importance of 3D otolith shape analysis for stock discrimination.