The reduction from three‐ to two‐dimensional analysis of the permeability of a fractured rock mass introduces errors in both the magnitude and direction of principal permeabilities. This error is numerically quantified for porous rock by comparing the equivalent permeability of three‐dimensional fracture networks with the values computed on arbitrarily extracted planar trace maps. A method to compute the full permeability tensor of three‐dimensional discrete fracture and matrix models is described. The method is based on the element‐wise averaging of pressure and flux, obtained from a finite element solution to the Laplace problem, and is validated against analytical expressions for periodic anisotropic porous media. For isotropic networks of power law size‐distributed fractures with length‐correlated aperture, two‐dimensional cut planes are shown to underestimate the magnitude of permeability by up to 3 orders of magnitude near the percolation threshold, approaching an average factor of deviation of 3 with increasing fracture density. At low‐fracture densities, percolation may occur in three dimensions but not in any of the two‐dimensional cut planes. Anisotropy of the equivalent permeability tensor varies accordingly and is more pronounced in two‐dimensional extractions. These results confirm that two‐dimensional analysis cannot be directly used as an approximation of three‐dimensional equivalent permeability. However, an alternative expression of the excluded area relates trace map fracture density to an equivalent three‐dimensional fracture density, yielding comparable minimum and maximum permeability. This formulation can be used to approximate three‐dimensional flow properties in cases where only two‐dimensional analysis is available.