The objective of this study was to determine the influence of solute properties and operational parameters on disinfection by-product (DBP) rejection by reverse osmosis (RO) and nanofiltration (NF) membranes. This was achieved by assessing the removal efficiency for 29 DBPs likely to be formed during disinfection of secondary effluents. The DBPs investigated were trihalomethanes, iodinated-trihalomethanes, haloacetonitriles, chloral hydrate, haloketones, halonitromethanes and haloacetamides. The performance of a NF and a low pressure RO membrane was investigated within a range of different pHs, temperatures, transmembrane fluxes, crossflow velocities and ionic strengths. Rejection decreased significantly with increasing temperature and decreasing transmembrane flux, while the influence of the other operational parameters was minimal with a few exceptions detailed in the manuscript. Multiple linear regression was used to determine the physico-chemical solute properties contributing significantly to DBP rejection. For NF, geometric parameters were revealed to be the dominant molecular descriptors influencing rejection, whereas for RO, besides size exclusion, solute-membrane interaction played an important role. A predictive model based on multiple linear regression was established that could forecast rejection of DBPs as a function of membrane operation parameters and DBP properties. (C) 2014 Elsevier B.V. All rights reserved.