Interest is growing in developing membrane bioreactors (MBRs) to replace ion exchange for nitrate removal from drinking water. However, few published studies have successfully managed to retain exogenous or biologically derived carbon. This study determined an optimum C:N by substrate breakthrough rather than maximum nitrate removal. By dosing <C:N 1.52, ethanol (electron donor) could be retained and nitrate removal up to 92% could be observed. Carbon limited control led to nitrite formation <0.35 mg NO2(-)-N, which was oxidised to <0.02 mg NO2(-)-NL(-1) by chlorination. Residual organics showed little competition with nitrite during short-term chlorination; thus chlorine demand could be predicted by the nitrite residual (5.0 +/- 0.1 mg Cl2 mg NO2(-)-N(-1)). Biomass was completely retained and low residual dissolved organic carbon (DOC) (0.4 +/- 0.7 mg L(-1)) indicated high rejection of influent DOC and soluble microbial products (SMPs). Residual DOC exhibited low reactivity with long-term chlorination (trihalomethane formation potential (THMFP) 72.7 +/- 7.7 microg L(-1)); however, the impact of substrate breakthrough on THMFP was significant. Immersed heterotrophic MBR produced consistent high quality product water at steady state, comparing favourably to previously trialled heterotrophic fixed film membrane and packed bed processes.