Abstract: Extension of the operational range of underwater location techniques has required incorporation of reliable acoustic propagation models, that closely predict oceans' behavior. While opening new possibilities, such as the use of focusing techniques in parallel with existing localization methods, this evolution also resulted in increased sensitivity to the accuracy of the models used. We show how a performance analysis tool based on information theoretic measures o] model resemblance can be used in this context in two distinct ways. The first enables the analysis of the sensitivity to the propagation model, and thus identifying the most critical model parameters, for which accurate estimates must be provided. In the second, we discuss how a priori knowledge of the model structure can be exploited to improve performance of focusing/localization methods.