A depth filtration model has been developed, based on the three-dimensional void network model Pore-Cor. The geometry of the void network is fitted, by means of an. eight-dimensional Boltzmann annealed amoeboid simplex, to the porosity and percolation characteristics of stainless steel sintered filters. measured by mercury porosimetry. Preferential and critical flow paths through the network are calculated. Particles from an experimental size distribution are fed along these flow-biased paths, and when straining occurs, the flow paths are re-calculated. We show that the model usefully reproduces experimental filtration efficiencies as a function of pressure drop, measured by single pass tests. We also offer a critique of current measurements of filtration efficiency, suggesting the use of a new "alpha efficiency" rather than the standard beta efficiency. The model is currently being adapted to accept porometry as well as porosimetry data, hence avoiding the use of mercury. (C) 2009 American Institute of Chemical Engineers AIChE J, 55: 3134-3144, 2009