The most common method of filtration is via fibrous nonwoven media. Fibrous filters are generally characterized by their collection efficiency and pressure drop. Traditional computational studies in this area are typically based On unrealistic 2-D geometries with the fibers simply placed in a lattice (regular array) perpendicular to the flow. The traditional approaches however. do not permit studying the relation between the 3-D structure of a filter media and its performance. In this study, for the first time. a virtual 3-D web is generated based on the fiber orientation information obtained from analyzing microscopic images of lightweight spun-bonded filter media. Pressure drop and collection efficiency of our virtual filter are simulated and compared with the previous 2-D analytical and numerical models as well as experiment. Our pressure drop calculation, unlike the previous models. showed a perfect agreement with the predictions of the Davies' empirical equation. The collection efficiencies obtained from simulating a thin spun-bonded filter media challenged with submicron particles having diameters ranging from 50 to 500 nm showed a similar trend as that of the previous 2-D models. For the solid volume fraction (SVF), filter thickness, and the fiber and particle diameters considered in this study. we found collection efficiencies higher than that of the above mentioned 2-D models with a relatively good agreement with experimental data obtained from a TSI 8130 filter tester. (c) 2006 Elsevier Ltd. All rights reserved.