Abstract: The latin hypercube sampling technique was used to obtain sample points of structure parameters for angular contact ball bearings. A combination method of the Powell algorithm and Newton-Raphson method was proposed for solving nonlinear equations and then the as-obtained the bearings stiffness were applied on fitting response surface functions based on the Kriging model. The influence of the mean and variation of structure parameters for angular contact ball bearings on its stiffness sensitivity was simulated using the Monte Carlo method. The results show that the mean and variation of structure parameters for angular contact ball bearings have an adverse effect on the sensitivity of the axial and radial stiffness. Bearings stiffness is most sensitive to the variation of ball diameters， then curvature radius of the inner and outer raceway groove takes the second place， and the outer raceway diameter is the least. This research may bring forward the theoretical basis for selecting the design parameters and machining accuracy of bearings as well as separation of parts.

Key words: angular contact ball bearing, stiffness, sensitivity, structure parameters, randomness