Several varieties of macadamia nuts are studied using X-ray microtomography, in order to determine a link between structural properties and storage stability. Fresh nuts-in-shell (NIS) of the predominantly grown hybrid varieties (A38, 816, 842, Daddow and 246) are scanned, the 3D tomographic reconstructions are then segmented, i.e. pixels labelled either as shell, void (inner and outer), and kernel. The automatic segmentation process consists in two steps: positioning of markers that identify each of these regions (by labelling pixels that are certain to belong to the region), and delimitation of these regions. The first step is performed with typical image processing operators (thresholding, connected component analysis, distance transform), while the second step uses the watershed algorithm. Once all regions are identified, pixel counting provides measures of volumes for each region, which, coupled with weighing, gives densities. We show that from physical quantities such as kernel volume percent or shell density, two main classes of nuts can be derived. A relationship between these observations and the behaviour of the nut varieties considered by this study during postharvest operations is discussed and further research work is proposed.