Abstract: The complex fracture network of rock mass provides the main paths for fluid transportation. The research on the anisotropic characteristics of fracture network from a quantitative way is an important and fundamental subject. Based on the statistical reconstruction of discrete fracture network and RJNS^3D, the quantitative description and anisotropic characteristics of fracture network are studied. The results show that: (1) The fracture density, fractal dimension, connectivity index and the relative dimensionless fracture density can quantitatively describe the randomness and connectivity of fracture network. (2) When the measuring window is perpendicular to the average orientation of fractures, a fracture network with a good connectivity can be generated and its quantitative indicates are obviously higher than other directions. (3) Compared with the average values of each index, the difference coefficient of the four indies reaches 0.111~0.493, 0.015~0.072, 0.007~0.180 and 0.008~0.724, respectively. (4) The fractal dimension is not the best index for the description of anisotropic characteristics, and choosing the multi-indicators or the relative dimensionless fracture density can be a good choice. The analysis results of the engineering application show that the method of the DFN reconstruction and measuring window can achieve the quantitative description of the anisotropic characteristics of fracture network. And it may provide some foundations for the study on the percolation characteristics of fractured rock mass.