More and more attention has been paid to the results of quantitative assessment of slope failure risk by using uncertainty analysis method. In view of the problem that only stationary random field was used to simulate the distribution characteristics of shear strength parameters in the existing risk assessment of slope failure, in this paper, taking an undrained saturated clay slope as an example, combined with finite element limit analysis, strength reduction method and Monte Carlo simulation, the influence of shear strength parameters of soil simulated by non-stationary random field model and stable random field model on Calculation of slope instability risk assessment results were discussed.The results show that the traditional method of calculating the risk of slope failure by using stationary random field will overestimate the risk of slope failure, which will increase the project cost.The slip surface of slope with high safety factor is not necessarily a shallow slip surface. The vertical correlation distance of slope has little influence on the calculation of its failure risk by using non-stationary random field, while the vertical correlation distance of slope has significant influence on the calculation of its failure risk by using stationary random field. The influence of vertical correlation distance on position and shape of sliding surface is weak.The research results can provide guidance for the accurate evaluation of slope stability.