A novel approach using adsorption pretreatment coupled with parallel factor analysis of excitation-emission matrices (PARAFAC-EEMs) was proposed to rapidly identify the role of fluorescent natural organic matter (NOM) fractions in ultrafiltration (UF) membrane fouling for given feed water and membrane. Two anion exchange resins, three polymeric resins and a powdered activated carbon (PAC) with diverse adsorption characteristics were adopted in the adsorption of raw water. The matrix of fluorescent NOM was substantially changed by the adsorption pretreatment, which rapidly enabled a diversity of fluorescent NOM with the components of an identical origin. PARAFAC-EEMs were employed to characterize the variation of fluorescent NOM during the adsorption, which in turn allowed to correlate the fluorescent NOM components with the membrane fouling in the following UF test. The results showed that, for the Songhua River water and PVDF UF membrane (Litree, China), three statistically significant fluorescent components (arbitrarily labeled C1, C2, andC3) were obtained through PARAFAC analysis, in which two humic-like (C1 and C2) and one protein-like (C3) components were identified. Total and irreversible fouling were highly correlated with the maximum fluorescence intensity (F-max) of C3, while poorly correlated with the F-max of C1 and C2, suggesting that C3 can act as an indicator for th e UF membrane fouling. Moreover, C1, C2 and C3 have a considerable synergistic contribution to the irreversible fouling of the UF membrane with the analysis of multiple linear regression. These results are expected to provide important implications for monitoring and controlling UF membrane fouling. (C) 2014 Elsevier B.V. All rights reserved.