A quantitative analysis of variations,in granule microstructure based upon changes in primary particle size and bed preparation is presented. The granule microstructures are obtained using X-Ray Computed Tomography (XRCT). An algorithm is developed to measure the number and size of macro-voids (pore space with volume equivalent size greater than or equal to 30 pm or 3 times the primary particle size). Four size fractions of alumina, ranging in primary particle size from 0.5 mu m to 108 mu m, are sieved using three different sieve sizes to create static powder beds from which single-droplet granules are produced. The analysis shows that large macro-voids exist in ultra-fine powders (0.1-10 mu m). The macro-voids take up to 7% of the granule volume and the largest macro voids are 200-700 mu m in volume equivalent size. Changing the sieve preparation changes the size and total volume of macro-voids. In contrast, there are very few macro-voids in granules formed from coarser powders. This study shows that micron sized powders have the opportunity to form complex structures during granulation and that the handling history of the materials should receive greater scrutiny than it currently gets. (C) 2016 Elsevier B.V. All rights reserved.