Ultrafiltration-induced polarization chromatography (UPC), belonging to the class of field-flow fractionations, is examined in some depth mathematically, and predictions based on this examination are tested against preliminary experimental data. Both analysis and experiments are performed for slit flow between porous confining walls with a superimposed cross flow perpendicular to these walls. Integration of the diffusion equation for an idealized two-dimensional flow suggests that UPC is a very promising alternate to gel permeation chromatography, both for determining polymer diffusion coefficients and estimating the molecular-weight distribution of mixtures. These predictions are borne out to a substantial degree by experiments with monodisperse native proteins, but it is clear that important practical problems remain to be solved. The most important of these are adsorption of protein and deviation of flow distribution from that modeled.