The performance of a 6 kW proton exchange membrane (PEM) electrolyzer was modeled using a semiempirical equation. Total cell voltage was represented as a sum of the Nernst voltage, activation overpotential and ohmic overpotential. A temperature and pressure dependent Nernst potential. derived from thermodynamic principles, was used to model the 20 cell PEM electrolyzer stack. The importance of including the temperature dependence of various model components is clearly demonstrated. The reversible potential without the pressure effect decreases with increasing temperature in a linear fashion. The exchange current densities at both the electrodes and the membrane conductivity were the coefficients of the semiempirical equation. An experimental system designed around a 6 kW PEM electrolyzer was used to obtain the current-voltage characteristics at different stack temperatures. A nonlinear curve fitting method was employed to determine the equation coefficients from the experimental current-voltage characteristics. The modeling results showed an increase in the anode and cathode exchange current densities with increasing electrolyzer stack temperature. The membrane conductivity was also increased with increasing temperature and was modeled as a function of temperature. The electrolyzer energy efficiencies at different temperatures were evaluated using temperature dependent higher healing value voltages instead of a fixed value of 1.48 V. (C) 2008 Elsevier B.V. All rights reserved.