Abstract: Particle size is a key technical index in ore grinding-classification process, which is so difficult to control effectively with conventional loop control strategies. An intelligent control setting approach based on CBR for the grinding-classification system is proposed by combining intelligent method with conventional loop control. Aiming at the interval control of the particle size index and according to the information on boundary conditions and current operating conditions, a model is set intelligently to update automatically all the basic loop setpoints, thus avoiding the subjectivity and randomicity arising from arbitrary process control setting. Then, the outputs from control loops can track down the updated setpoints relevantly to control the particle size within the target zone. The approach proposed has successfully been applied to the grinding-classification process of a hematite ore processing plant, and its effectiveness is proved evidently.