The first step of fuel cell stack characterization consists in experimental testing of its global electrochemical performances under various working conditions. The second step is devoted to link the collected data to local species concentrations, membrane hydration and resulting current density in the stack. In the project PICOS, we have to characterize a 1.5 kW 12-cells' stack of 330 cm(2). The stack performances are measured on a specific test bench and analyzed using a fuel cell-stack modeling tool. The test bench allows to control the oxygen (or air) and hydrogen flow rates and humidity at the inlet of the stack, the pressure at the outlet, and the delivered current. Each cell potential and the total stack potential are monitored, the produced water is condensed at the outlet in order to measure its amount. The model used is a mono-dimensional description of a single cell. It is based on a semi-empirical law for the local electrochemical response computation. The local pressures, temperatures, gas compositions and membrane hydration are computed dynamically. A semi-empirical electrochemical model of the active layer is fitted to experimental data, the gas diffusion layer porosity and electro-osmosis coefficient of the membrane have been adjusted to obtain the best fit. The simulations calculate the cell potential and the current density profile in the cell for various feeding conditions. The simulations allow us to explain the effect of gas hydration on the stack potential and to predict possible drying out of the membrane. (c) 2005 Elsevier B.V. All rights reserved.