Computing flows and phase transport in porous media requires a physically representative geometric model. We present a simple method of digitizing the structure of fibrous porous media commonly used in polymer electrolyte membrane (PEM) fuel cells, the so-called gas diffusion layer (GDL). Employing an inverted microscope and image recognition software we process images of the GDL surface collected manually at different focal lengths with micrometer accuracy. Processing the series of images allows retrieval of local depths of the salient in-focus structural elements in each of the different images. These elements are then recombined into a depth-map representing the three-dimensional structure of the GDL surface. Superimposition of the in-focus portions of the structural elements distributed throughout the stack of images yields digitized data describing the geometry and structural attributes of the 3 D surface of the GDL fibrous material. (C) 2009 Elsevier B.V. All rights reserved.