Concepts for the effective MWCO of tight-UF membranes, and apparent diffusion coefficients for NOM, were introduced to determine the mechanisms influencing NOM removal and to explain the various behaviors of NOM removal by UF membranes with different hydrophobicities, permeability, and surface charges. Colloidal NOM (COM) and non-colloidal hydrophobic NOM (NCD HP) constituents were chosen for the evaluation of two different UF membranes. For a relatively hydrophobic, relatively high permeability, and less negatively charged UF membrane, the hydrophobic fractions of COM were preferentially removed and were also removed by a size exclusion mechanism (i.e., both hydrophobic interaction and size exclusion mechanisms). The NCD HP exhibited no such preferential removal of the hydrophobic fractions, but could be removed by a size exclusion mechanism (i.e., only size exclusion mechanism). With a relatively hydrophilic, relatively low permeability, and more negatively charged UF membrane, COM exhibited no preferential removal of the hydrophobic fractions, but could be removed by a size exclusion mechanism (i.e., only size exclusion mechanism). Whereas the hydrophobic fractions of the NCD HP were preferentially removed, these could not be removed by a size exclusion mechanism (i.e., only hydrophobic interaction mechanism). The apparent diffusion coefficients of NOM, as determined from NOM diffusion experiments using a diffusion cell equipped with a regenerated cellulose membrane, were much lower than those calculated by the Stokes-Einstein relation. The diffusion coefficient of NOM is expected to be used to predict and explain NOM transport behaviors in tight-UF membranes.