Rotary drums are commonly used in particulate solid industries for mixing, coating and reactions. The process is often accompanied by undesirable breakage of granules. For this reason, a scaled-down version is sometimes used as an attrition testing device. In this work, the attrition of granules inside a rotary drum at 18, 35 and 52 rpm drum rotation speeds for 4000 cycles is studied. The granules used in this study have been produced by extrusion and spheronisation with a size range of 500 to 1000 pm. The rotary drum has an internal diameter of 0.39 m, axial length of 0.3 m and a single baffle. The extent of breakage is quantified by sieving out fine debris which is two sieve sizes smaller than the feed particles. To relate the extent of breakage in the drum to granule characteristics, single granule impact tests have been performed on one type of granule at several velocities. The effects of particle size and impact velocity are analysed and a power-law relationship is fitted between impact velocity and single granule breakage. This information is then used to simulate granule breakage in a rotary drum by Distinct Element Method (DEM). The drum is simulated for 5 rotations at the rotational speeds stated above and the breakage rate is extrapolated to 4000 cycles where it is compared to experimental results obtained. The trends for particle breakage in both experiments (determined by sieving) and extrapolated DEM simulations are in agreement however the orders of magnitudes are different. The comparison shows that the extent of breakage obtained from extrapolated simulations is overestimated at drum speed of 35 and 52 rpm and underestimated at 18 rpm. There is close agreement between experiments and extrapolated DEM simulations for particle breakage at 18 rpm only after 4000. Furthermore, the effect of air drag on the attrition of granules by impact at a drum rotation speed of 52 rpm is investigated, where it is found to significantly reduce the breakage results. (C) 2011 Elsevier B.V. All rights reserved.