The removal of microcystins (MC) by powdered activated carbon adsorption/ultrafiltration (PAC/UF) was investigated, focusing on PAC dose and addition mode, MC initial concentration (expressed as MC-LReq) and on the impact of background natural organic matter (NOM), assessed through model compounds (a mixture of tannic and humic acids) and Microcystis aeruginosa culture. Constant flow experiments were performed with a hydrophilic UF hollow-fibre membrane and a mesoporous fine-powdered activated carbon. In the absence of background NOM, PAC/UF with 10 mg/L PAC and up to 20 mu g/LMC-LReq feed concentration achieved 93-98% MC removal and a cycle-averaged permeate concentration below the drinking water guideline-value adopted by the World Health Organisation (WHO) for microcystin-LR variant. Single-pulse PAC dosing in the beginning of the UF-cycle allowed slightly lower MC concentration in the permeate compared to the multi-pulse PAC addition whereas no differences were found in terms of transmembrane pressure. Hydraulic retention time of 34 and 55 min resulted in similar permeate quality. NOM type and concentration and MC initial concentration determined the PAC dose to be used. While 10 mg/L PAC effectively controlled ca. 5 mu g/L MC-LReq in a model water with 2.5 mg/L NOM or with M. aeruginosa culture (cells and algogenic organic matter) 15 mg/L PAC were unable to achieve the WHO quality with a water containing higher concentrations of NOM (5 mg/L) and microcystins (ca. 20 mu g/L MC-LReq). (C) 2009 Elsevier B.V. All rights reserved.