In mining, the particle size obtained after grinding greatly influences the amount of mineral recovered, so to obtain an efficient operation with a reduction in production cost, the grinding circuit must have optimal control, however, the system is complex and has many input and output variables that interact with each other. This paper uses a 4x4 MIMO model that has been obtained by an experimental process and implements a model-based predictive control strategy (MPC) in the generalized predictive control (GPC) modality, with optimal control characteristics. The objective is to investigate the effect that occurs on the control due to degradations of the components of the grinding process, which is reflected in changes in the delays and the sensitivity of the transfer functions. The effects on control are evaluated in parameters such as overshoot, rise time, and disturbances in controlled variables caused by variations in delays and amplitudes in the transfer functions of the model. The effects presented and analyzed in the results section provide considerations that are of interest for tuning the control system, it also provides important criteria for plant operation and maintenance. The results have been obtained by simulation with Matlab.