ABSTRACT: This contribution takes advantage of three multiscale imaging techniques to predict permeability from the actual 3D percolating pore structure of a natural claystone, planned in France for long term nuclear waste repository. These imaging techniques (micro-tomography i.e. micro-CT, Focused Ion Beam/Scanning Electron Microscopy i.e. FIB/SEM and Transmission Electron Microscopy i.e. TEM) have proven necessary to quantify percolating nanoscopic and mesoscopic porosities (typical sizes of a few nm to a few tens of nm). The originality of this contribution is to predict permeability in 3D from the nanoscale (with data from TEM), to the mesoscale (with data from FIB/SEM) and up to the macroscale (with micro-CT data). No fluid/solid interaction is accounted for. While nanoscale and mesoscale images benefit from the Lattice-Botzmann Method to solve the Stokes equations, the Darcy equation at the macroscale is solved by a finite volume discretization technique coupled to conjugate gradient. Good agreement with experimental permeability data for undisturbed claystone is achieved when non percolating mesoscale pores are coupled to percolating nanometric ones. This combination (percolating nanoscopic pores + non percolating mesoscopic pores) drives fluid transport in natural claystone.