Constrictions between voids control the filtration and permeability properties of granular materials. This study uses high-resolution microcomputed tomography images and discrete-element modelling to analyse two important characteristics of constrictions in granular filters: (a) the constriction size distribution (CSD) and (b) the constriction density per unit volume. The results demonstrate the importance of the particle size distribution (PSD) and void ratio of the granular material in determining the constriction density, with more widely graded materials having more densely spaced constrictions. The PSD is shown to be the main determinant of the CSD, in agreement with previous studies. The data are used to examine proposed approaches to estimate constriction spacing or void size.