Water quality results related to disinfection by products (DBPs) from a bench-scale study designed to evaluate surface water nanofiltration (NF) for the City of Houston are analyzed to provide insights into mechanisms of DBP formation and control. One contribution of this manuscript is that we report on DBP speciation following NF of Lake Houston water containing very low concentrations of the bromide ion (< similar to 50 mu g/L or 0.6 mu M). Rejection of natural organic matter (NOM) and DBP precursors decreased with increasing recovery indicating diffusion-limited transport of these macromolecular contaminants under conditions typical of drinking water NF. Additional evidence of diffusivity effects on transport was provided by decreasing permeation coefficients of polyethylene glycols with molecular weight. Haloacetic acid (HAA) molar concentrations were higher than corresponding trihalomethane (THM) concentrations in NF permeates contrary to expectations from the DBP rule where lower mass-based maximum contaminant levels are enforced for HAAs. Similar to conventional treatment, ultraviolet absorbance at 254 nm (UV254), dissolved organic carbon (DOC), and chlorine consumption were found to be excellent surrogates for aqueous DBP concentrations even in chlorinated nanofiltered waters. Mole fractions of mono-, di-, and trihalogenated HAA species were invariant with respect to changes in Br-/DOC molar ratio suggesting that the brominated HAAs are formed through the same mechanisms as the chlorinated ones in NF permeate waters. Additionally, bromine incorporation into both THMs and HAAs increased with increasing Br/DOC ratio suggesting that the brominated THM and HAA species form first consuming the reactive precursor sites and in the process restricting formation of the chlorinated species. Bromide utilization was observed to decrease with Br-/DOC molar ratio demonstrating that nanofilter permeate waters were substantially limited with respect to NOM and DBP precursors. (C) 2008 Elsevier B.V. All rights reserved.