A new simulator for flow of aqueous suspensions and deposition of non-Brownian particles in granular media can predict the pattern of deposition and concomitant reduction in permeability as functions of depth, time and system parameters. The porous structure of the granular medium represented as a 3-D network of constricted pores considers the converging - diverging character of flow within pores. Using Lagrangian-type simulation tile particle deposition rate was calculated. Gravity and dr ag, as well as hydrodynamic and physicochemical interactions between suspended particles and pore walls, were considered in calculating 3-D particle trajectories. Deposit configurations were computed and the evolution of the pore structure was simulated at discrete time steps. Changes in the pore geometry and nature of the collector surface affect flow and trajectory computations directly. Clusters of deposited particles were allowed to become reentrained if exposed to shear stress higher than a critical value. Reentrained clusters, which moved through downstream por es, might redeposit downstream at suitable sites and cause clogging of sufficiently narrow pores. Particle clusters clogging pores have a finite permeability, which significantly affects the system's transient behavior Clogged por es act as collectors of solitary particles and of reentrained clusters, and substantially affect tile transient behavior of the filter The loss of permeability was monitored by calculating pore and network hydraulic conductance at each time step. Numerical results for the loss of permeability, temporal evolution of filter efficiency, and specific deposit profiles are based on suspension flow simulations in a typical granular porous medium.