Abstract: Three-dimensional morphology of the macroscopic undulation component and asperities of rock joints can directly affect the shear mechanical properties. Understanding the two-order roughness anisotropy features of joint surfaces at different sampling intervals is important to recognize the shear failure mechanism of rock mass. In view of the fact that there are few studies on the anisotropy characteristics of two-order roughness of rock joints at present, the two-order roughness components based on limit sampling interval morphological classification method are obtained, and the anisotropy coefficient RAC^3Dconsidering the dominant and disadvantageous directions of roughness is proposed. The anisotropy coefficients of the two-order roughness of three joint samples at different sampling intervals based on point cloud sparse method are calculated. The results show that the anisotropy coefficient of the second-order roughness first increases and then rapidly decreases with sampling interval increase, and has a good quadratic polynomial fitting relationship with sampling interval. The interval effect of roughness anisotropy coefficients on first-order roughness is not obvious at a small sampling interval. When the interval exceeds the limit sampling interval, anisotropy coefficient decays rapidly and has a negative numerical power function with the sampling interval. The analysis of the two-order roughness morphology geometry of the rock joints shows that the speed of the roughness attenuation rate in the advantage and weakness directions with the sampling interval increase is the key factor that causes the variation of the anisotropy coefficients of two-order roughness. The more obvious the anisotropy characteristics of joint surfaces roughness are, the more influenced they are by the variation of sampling interval. This paper can provide a reference for the study of the sampling interval effect on the two-order roughness anisotropic features of rock joints.