An investigation of the anisotropic permeability of a carbon cloth gas diffusion layer (GDL) based on the integration of X-ray micro-tomography and lattice Boltzmann (LB) simulation is presented. The method involves the generation of a 3D digital model of a carbon cloth GDL as manufactured using X-ray shadow images acquired through X-ray micro-tomography at a resolution of 1.74 µm. The resulting 3D model is then split into 21 volumes and integrated with a LB single-phase numerical solver in order to predict three orthogonal permeability tensors when a pressure difference is prescribed in the through-plane direction. The 21 regions exhibit porosity values in the range of 0.910–0.955, while the average fibre diameter is 4 µm. The results demonstrate that the simulated through-plane permeability is about four times higher than the in-plane permeability for the sample imaged and that the corresponding degrees of anisotropy for the two orthogonal off-principal directions are 0.22 and 0.27. The results reveal that flow channelling can play an important role in gas transport through the GDL structure due to the non-homogeneous porosity distribution through the material. The simulated results are also applied to generate a parametric coefficient for the Kozeny–Carman (KC) method of determining permeability. The current research reveals that by applying the X-ray tomography and LB techniques in a complementary manner, there is a strong potential to gain a deeper understanding of the microscopic fluidic phenomenon in representative models of porous fuel cell structures and how this can influence macroscopic transport characteristics which govern fuel cell performance. Copyright © 2010 John Wiley & Sons, Ltd.