Abstract: To make the two-dimensional roughness parameters fully reflect the directional features of rock joints,and overcome the influence of sampling interval on the roughness parameter,the joint surface is divided into a series of rectangular-shaped asperities,and a computational model for determining the deduced height of a rectangular-shaped asperity is proposed. A new parameter c for describing surface roughness is proposed by calculating the average value of the deduced height of rectangular-shaped asperities in the same shear direction. Index c is obtained by combining the two failure modes of rectangular-shaped asperity to determine the deduced height. It can combine the geometric characteristics of joints with the shear strength,and then combine the roughness of joints with the shear strength. This provides a possibility for the peak shear strength model of rock joints considering the shear direction. The parameter c is able to objectively reflect the roughness and show the directional characteristics of the shear strength of rock joints. Taking the standard JRC profile as an example,the calculation process of the roughness parameter c is expounded. The obtained parameter c under a certain measurement interval has the drawback that the roughness with smaller measure- ment interval will be concealed,and the contribution of small-scale roughness to shear resistance should not be neglec- ted. The parameter c with different sampling interval is obtained,and the result shows that the parameter c decreases with the increase of sampling interval and shows a fractal feature. Based on the fractal theory,the roughness of joints can be described by two fractal parameters,and these two parameters are independent of measurement scale. The rela- tionship between the roughness parameters and JRC is proposed. It is proved that the new method can establish a good relationship with the shear strength of rock joints.