Membrane filters are often used in the semiconductor industry to remove small particles from liquids. Filters have recently been developed that use adsorption as the dominant particle capture mechanism by creating a surface that is electrostatically neutral in the fluid of interest. Consequently, these filters provide high permeability as compared to traditional filters with equivalent particle retention because they do not rely solely on particle sieving. This paper describes efforts to elucidate the mechanistic relationship between the relevant membrane, particle and fluid properties that control retention in an adsorptive membrane. Experiments and modeling were focused on aqueous streams of varying pH and ionic strength. A network model was developed to describe the fluid (and particle) flow within the membrane on a microscopic level. This network model was combined with a DLVO-based particle adhesion model to form a complete representation of a membrane filtration process. Comparing this retention model with experimental results demonstrates the utility of this tool in predicting the effect of particle size, pore size and ionic strength on particle retention. (C) 2004 Elsevier B.V. All rights reserved.