In this study, natural organic matter (NOM) in source water, as well as the treated water after coagulation with or without potassium permanganate (KMnO₄) preoxidation, was characterized by using high performance size exclusion chromatography with organic carbon detector (HPSEC-OCD) and fluorescence excitation emission matrices (F-EEMs) with parallel factor (PARAFAC) analysis. Bulk parameters, such as dissolved organic carbon (DOC) and ultraviolet light absorbance at 254 nm (UV₂₅₄), were also analyzed. The results show that KMnO₄ preoxidation caused the breakdown of high molecular weight (MW) organics into low MW organics. All organics, whether those that existed in the source water or those generated by KMnO₄ preoxidation, could be partly removed by coagulation. Combining the derived organic fractions obtained from HPSEC-OCD with peak-fitting and from F-EEMs with PARAFAC on the same sample, humic substances have been specified as the main organic composition. Further, the predictive models for trihalomethanes formation potential (THMFP) and haloacetic acids formation potential (HAAFP) based on organic fractions from HPSEC-OCD have higher accuracy than those based on the components from PARAFAC modeling. These models provide useful tools to specify the organic fractions from HPSEC-OCD and F-EEMs that constitute active precursors towards trihalomethanes (THMs) or haloacetic acids (HAAs) formation in water. Further, by knowing the major organic precursors, it would facilitate choosing the appropriate water treatment process for disinfection by-products (DBPs) control.