This work is divided into two main sections. First, a review of the most frequently used approaches to molten carbonate fuel cell modelling is presented. The models available in literature are found to be classified in three different groups which are carefully described in the paper. Then the models are compared and their ability to predict fuel cell performance far from the design point is tested. It is concluded that even though some models seem to be accurate close to a reference point, their estimates differ largely far from these working conditions. Accordingly, a model is selected amongst those previously reviewed and its capacity to replicate real fuel cell performance is checked against experimental data available in open literature. Such validation confirms that the model is fully satisfactory as it is able to predict fuel cell performance within less than 1.0-1.5% deviation for a wide variety of test setups. This remarkable accuracy is further reinforced by the fact that the model does not rely on experimental information (known a priori) about the system being modelled. The work also introduces an interesting approach to estimate average gas compositions to be used when evaluating Nernst potential. This approach is based on arithmetic and logarithmic means for cathode and anode respectively and it proves to yield accurate results. Overall, the work provides researchers willing to construct their own models of performance with a valuable guide to assess them in such process. Copyright (C) 2011, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights reserved.