Abstract: A discrete mathematical model based on random linear Kalman fihering is introduced to eliminate random disturbance noise in the process of GPS slope deformation monitoring and to improve the validity of monitoring data. On the base of GPS slope monitoring data in Shuichang Iron Mine, the filtering value of deformation and the velocity of displacement at each point in each stage can be calculated with the mathematical model and the slope deformation at each point in the next stage can be estimated and predicted. It is proved with an example that the deformation obtained by Kalman filtering is more approximate to the real slope deformation.