ABSTRACTIn this work, an indirect inverse method utilizing sensitivity analysis is employed to help understand the reasons for model insensitivity. The results of the sensitivity analysis allow the modeler to delineate insensitive areas of the model where inverse procedures will be more subject to error. Sensitivity coefficients are defined and discussed. A differential equation is developed for the sensitivity coefficients that will generally be solved by numerical techniques. A relatively simple least squares' inverse procedure is used on a hypothetical model to illustrate typical problems that can be encountered. In particular, the effect of data accuracy is considered. The low sensitivity areas of models are generally related to small values of δh/δx and δh/δt. The fact that considerable error in the transmissivity and storativity may occur in areas of low sensitivity should not be looked upon as a failing of the inverse procedure. It is simply a fact that not all areas of the model have been stressed equally. The main advantage of the present work is that areas of low sensitivity may be delineated.