It has been hypothesized that a set of simple tests can be used to characterize ("fingerprint") source feed waters and membranes such that integrating feed solution and membrane properties can lead to optimal membrane selection and operation. Knowledge of source water characteristics is important because it can aid in the proper selection of a membrane that best resists fouling and exhibits good flux recovery after cleaning to optimize the operation of membrane treatment facilities. Developing a fingerprinting technique to gain knowledge of source water characteristics and applying it to a predictive model was the focus of this research. Flow Field Flow Fractionation (Fl-FFF) is one of several such tests that can provide "fingerprint" information about source water quality that can be integrated with known membrane properties and correlated to flux decline in cross-flow membrane filtration. Flow FFF is typically used to separate solutes based on size and is commonly described as a single-phase chromatographic technique. The flux decline potential of a solution is related to the concentration of the solutes contained in the mixture at the membrane surface, their potential for forming a boundary layer or cake-like mass at that concentration, and their potential for irreversibly adhering to the membrane material at the surface and/or within pores. An ideal model to predict membrane flux should be based on first principles. However, such models are not tractable for the complex mixtures that are real world water supplies. Means to "collapse" the behavior of a complex mixture into a single "effective" medium or component would greatly simplify the task of predicting membrane filtration performance. Several of the qualitative relationships that define a solution's potential for causing flux decline relate to a variety of the physico-chemical properties of the solutes. Since many of the same properties govern the Fl-FFF methodology, it may be a useful tool for defining the physico-chemical properties of an effective medium that is being filtered, and results can thus be interpreted in terms of the parameters in a flux decline model. The general approach for this research was to perform Flow FFF measurements on combinations of silica colloids (dp = 74 nm) and whey protein (MWavg = 25,500 Da) under different cross field velocities and solution ionic strength conditions. The experimentally obtained retention ratios and hydrodynamic conditions were used to provide input parameters to an advection-dispersion transport model solved for the Fl- FFF system. The residence time distributions (RTDs) were transformed to a residence diameter distribution function (RDDF). A moment analysis of experimental ("real") vs. model ("ideal") RDDFs was used to quantify differences. Includes 7 references, tables, figures.