Discrete element modelling (DEM) of particle flows inside ball mills involves following the trajectories and spins of all the particles and objects in the system and predicting their interactions with other particles and with the mill. It is necessary to simulate particles of many different sizes and densities interacting with complex shaped lifters and liner. The key ingredients are a fast and robust algorithm to predict collisions, a good collision model and an efficient and powerful method for describing the mill. Particle flows in a 5 m diameter ball mill are presented. The charge behaviour, torque and power draw are analysed for a range of rotation rates from 50 to 130% of the critical speed. Sensitivity of the results to the choice of friction and restitution coefficients and to the particle size distribution are examined. Segregation is an important issue in rotating mills. Size segregation of steel balls and size separation of small rocks from the larger balls are examined. Predictions of liner wear rates and distributions are made. Evolution of the liner profile can be modelled in order to predict the lifter life cycle and its effect on the mill operation. Collisional force distributions can be used as indicators of breakage and attrition. Such quantitative predictions, once validated, can be used as part of a program to optimise mill design and operation.