Predicting the macroscopic properties of porous media used in treatment processes is a complex task regarding 3-D structures at micro-, meso- and macro-levels. Currently, information is scarce concerning the influence, at a microscopic level, of the 3-D structure of fibrous media on the physical laws governing their macroscopic behaviour. Nevertheless, the relationship between macroscopic properties (pressure drop, treatment efficiency) and microstructure can be assessed thanks to suitable structure modelling theories. In this context, the present study proposes and compares different methods (mercury porosimetry and image analysis) for the structure characterization at a microscopic level Of filtering fibrous media, such as nonwoven and woven fabrics. The results obtained show a porous structure gradient in the thickness of the nonwoven media studied in terms of porosity, pore size and tortuosity factor. Moreover, the influence on structural parameters of media compression, when submitted to friction forces exerted by flow during filtration tests, is established. A model for the determination of multi-level pore size distributions from mercury porosimetry data is proposed. The "equivalent pore" model is used to estimate the tortuosity factor. The influence of measured structural parameters on fibrous media permeability is studied in a classical model for flow through fibrous media.