Constant pressure (CP) testing systems are often used in characterization of gas separation membranes. Unlike constant volume (CV) systems, the membrane tested a CP system is exposed to atmosphere via a flow metering device. Consequently, as long as the permeating gas has a composition different from that of the atmosphere, there is a driving force for diffusion of atmospheric gases towards the membrane. If as a consequence this there is a nonzero partial pressure of the back diffusing gases at the permeate side of the membrane they may permeate to the feed side of the membrane (back permeation). This paper presents a mathematical model that allows estimation of an error arising from back diffusion and back permeation in CP systems. The model is derived from the first principles assuming that there are no interactions between the forward and back permeating gases with each other and with the membrane. The theoretical predictions are then compared with experimental results obtained in a specially designed, fully-automated CP system. using poly-2,6-dimethyl-1,4-phenylene oxide (PPO) films in single gas permeation tests involving nitrogen and oxygen. The experimental results confirm theoretically predicted trends resulting from the phenomena of back diffusion and back permeation. However, the influence of these phenomena on the experimentally determined permeability coefficients is greater than that predicted by the model. (C) 2008 Elsevier B.V. All rights reserved.