This study evaluates methods used for granule shape characterization. The aim is to identify an optimal combination of shape factors to measure granule shape and roughness. Granules were prepared from microcrystalline cellulose (MCC), alpha-lactose, microfine cellulose (MFC), and dextrin, using a small-scale high-shear mixer. Granule shapes were analysed by measuring weight, by microscopic analysis, and by sedimentation analysis in olive oil. A comparison was made among (1) the aspect ratio, (2) the circularity, (3) a new projection shape factor defined in this paper, (4) the shape factor e(R), (5) the radial shape factor, (6) One-Plane-Critical-Stability (OPCS), (7) the Stokes' shape factor, and (8) a new mass shape factor defined in this paper. Besides the evaluation of the shape factors, (9) a roughness factor R is defined. When the shape of granules is evaluated, there can be effects of both shape and surface roughness. The scale of scrutiny of the measurement determines whether or not a shape factor is affected by surface roughness. We chose to call a protrusion smaller than 60 mum a roughness, and a larger protrusion a shape effect. The best evaluation of granule shape is obtained when shape and roughness are described by different parameters. The aspect ratio, the mass shape factor, OPCS, e(R) and the radial shape factor could not significantly discriminate between some of the different shapes. The other three shape factors give distinct differences in value for all different shapes used in this study. Of these factors, Stokes' shape factor was rejected because of the high viscosity of the oil needed, which results in a significant temperature dependency. Circularity and the new projection shape factor both work very well for our granules. The combination of the roughness factor with the new projection shape factor will provide a good description of granule shape and roughness. Granule shapes were(C) 2004 Elsevier B.V. All rights reserved.