Previous studies have shown that the absolute membrane cut-off and the value quoted by the manufacturers may be very different because of differences in methodologies and testing conditions. The origin of this discrepancy is often difficult to identify as the characterization method used by the membrane manufacturers is not specified. The goal of this study is to evaluate in terms of sensitivity, repeatability and reproducibility the methodologies used by a working group including membrane manufacturers, end-users and a research laboratory. The five selected membranes are hollow fibers used in drinking water production. They are made of various materials with different configurations (internal-external and external-internal type) and molecular weight cut-offs. Different types of tracers are used: dextrans, poly (ethylene glycol)s, poly(ethylene oxide)s, MS2 bacteriophage filtrated in single sized solution or in mixture. Results collected lead to a better understanding of the origin of discrepancies and allow to define the best operating conditions (tracer's type, working conditions range, data treatment methods,...) providing the most appropriate, accurate and reproducible testing protocol according to the selected application. (C) 2009 Elsevier B.V. All rights reserved.