The reliability of cellulose hydrolysis models is studied using the NREL model. An identifiability analysis revealed that only 6 out of 26 parameters are identifiable from the available data (typical hydrolysis experiments). Attempting to identify a higher number of parameters (as done in the original NREL model publication) results in significant errors on the parameter estimates. The reasons for this poor identifiability are related to (i) model structure complexity, inherently containing correlated parameters due to Michaelis-Menten type kinetics, and (ii) the available data, which are not informative enough (sensitivities of 16 parameters were insignificant). This indicates that the NREL model has severe parameter uncertainty, likely to be the case for other hydrolysis models as well since similar kinetic expressions are used. To overcome this impasse, we have used the Monte Carlo procedure to analyze the uncertainty of model predictions. This allows judging the fitness of the model to the purpose under uncertainty. Hence we recommend uncertainty analysis as a proactive solution when faced with model uncertainty, which is the case for biofuel process development research. (C) 2010 Elsevier Ltd. All rights reserved.