A three-dimensional, two-phase and non-isothermal numerical model has been developed in Part I of this two-article work. In this article, the parameter sensitivity analysis is performed. The influence of 11 major parameters, including the transfer coefficient, exchange current density multiplied by specific area, porosity, diffusion coefficient, absolute permeability and membrane phase conductivity, were investigated on the effect of the global polarization curve. The results show that the PEMFC global polarization curve is influenced by many parameters and the cathode-side parameters are a stronger influence than those on the anode side. Two different groups of parameters are provided which can result in almost the same global polarization curve, showing that the global polarization curve is not sufficient for the validation. Detailed discussion on the PEMFC model validation is conducted, and it is shown by numerical results that the global polarization curve plus the local current density distribution is still not sufficient for the model validation. A three-step validation approach is then proposed which can be expected to give a unique validation. The three steps are: validation of the global polarization curve; validation of the local current density distribution curve and validation of the cathode overpotential versus current density curve. Four further suggestions are proposed in order to solve the validation issue completely. These include the completeness of the data provided, the accumulation of benchmark data and the necessity for introduction of uncertainty analysis. (c) 2006 Elsevier B.V. All rights reserved.