While the scale-up from flat-sheet lab-scale experiments to larger modules for an industrial application requires consideration of the module geometry and according hydrodynamics, it is of high importance that the lab-scale experiments provide an accurate estimate of the membrane performance at ideal hydrodynamic conditions to make a proper selection and potentially allow for a model-based evaluation of the process performance. Due to inhomogeneity of the membrane material, it is however well known that the membrane flux determined by small samples in lab-scale experiments may vary in a considerable margin. The current study investigates these variations with an extensive experimental screening of lab-scale samples as well as a spiral-wound module for a commercially available polymeric membrane and pure solvent flux. This allows for an analysis of the comparability, as well as a best case estimate for the variance of the solvent flux, due to the industrially optimized membrane production process. The experimental results confirm considerable variations for single lab-scale samples, but also indicate good comparability on the basis of a statistical evaluation of the larger set of lab-scale samples. Based on the evaluation of the uncertainty of the solvent flux estimate, a minimum number of samples for an accurate estimate is determined and the recommendation is validated for additional membrane-solvent combinations. The importance of the derived uncertainty information for membrane selection and model-based conceptual process design is furthermore illustrated for a number of literature case studies. (C) 2019 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.