The liquid displacement method is a commonly used method to determine the pore size distribution of micro- and ultrafiltration membranes. One of the assumptions for the calculation of the pore sizes is that the pores are parallel and thus are not interconnected. To show that the estimated pore size distribution is affected if this assumption is not satisfied, we developed two models: (1) a model describing theflow through an isotropic porous membrane with uniform pores, and (2) a two-layer model for uniform pores that approximates the first model if the membrane thickness is larger than 10 times the pore radius. In the two-layer model the membrane skin layer is divided into two parts: the unconnected pore layer and a sublayer. This model is extended to describe pore activation as a,function of pressure with a pore size distribution in the unconnected pore layer (that is, membrane surface). It is shown that, depending on the membrane thickness or the sublayer resistance, the transmembrane pressure needs to be much larger than the critical pressure of the pores, to activate all the pores. If the sublayer resistance is over 10% of the resistance of the unconnected pore layer, the pore size is underestimated with the liquid displacement method; thus the number of pores is overestimated. Because the sublayer resistance is always larger than the unconnected pore layer resistance in an isotropic membrane with interconnected pores, we conclude that the estimated pore size distribution is always shifted toward smaller pore sizes than they really are. To use the liquid displacement method correctly, we suggest either counting the number of (active) pores or measuring the flux-pressure curve several times, while covering each time a different fraction of the membrane surface. (C) 2004 American Institute of Chemical Engineers.