Ball mill media sizing investigations were carried out for the ball mill circuits grinding minus 12.5 mm crushing plant product at Lundin’s Eagle Mine base metal concentrator in Humboldt, Michigan, USA. Batch grinding tests were carried out on mill feed with a torque-metered pilot mill with the same ball charge sizing (76 mm top size) as used in the plant. Calculated grinding rates of the coarsest particles were approximately four times higher for the plant mill than the pilot mill. One explanation is that coarser particles are retained in the continuous plant mill longer and thus exposed to more energy. This challenged the assumption used in population balance modelling that particles of all sizes, as well as the liquid, have the same residence time, as characterised by liquid tracer tests. A literature review revealed that this assumption is false, and that residence time as a function of particle size is not able to be determined precisely. To explore this further, a plant test was carried out by dosing a mill with 16–22 mm particles, coarser than any in the normal circuit feed, in conjunction with liquid residence tracing using salt and a conductivity probe. The test showed that the coarse particles were retained approximately twice as long as the liquid. It also showed that abrasion of these particles is contributing significantly to their size reduction.