Abstract: By statistical analysis of previous studies and laboratory tests on scale effects of rockfill materials, it could be argued thatthe size effect of particle strength should be the major cause for the scale effect obtained in the condition of the same particle source and the same sample preparation method. Then, in this study an expressionfor describing the strength of particles with different sizes was integrated into a survival probability model based on the Weibull distributions, and the size effect of particle strength was calculated using a combined finite-discrete element method (FDEM). The mechanism of scale effect was also examined at meso-scopic level. The simulation results show that the scale effect was dominated by anisotropy in normal contact force. With the same size effect parameter of particle strength, peak internal friction angle was decreased with the increasein the maximum particle size, while volume strain at damage point took a reverse trend. For the tests of same particle size, a decrease in the size effect parameter led to a decrease in peak internal friction angle but an increase in volume strain. When the size effect of particle strength was considered, increasing the maximum particle size caused the particle strength to decrease and the average particle normal and tangential contact forcesto increase. This explains why particles with different sizes show an obvious scale effect in laboratory tests.