Techniques of nonlinear regression were applied to estimate the hydrogeologic parameters (values of transmissivity or hydraulic conductivity), recharge, discharge, and boundary fluxes for steady state groundwater flow models of two field areas: Truckee Meadows, Nevada, and a cross section in the Hula Basin, Israel. Statistical techniques were used to estimate the degree of nonlinearity of the models, the goodness of fit of the models to field data, and the reliability of predictions to be made with the models. Goodness of fit was determined by analyzing the reliability and significance of computed parameters and the reliability of the computed head distribution. It was found that both of the field cases analyzed behaved at least approximately linearly based on Beale's measure of nonlinearity; that variable reliability of observed head data had to be incorporated into the analysis for Truckee Meadows; that residuals, transformed to eliminate heteroskedasticity and intercorrelation, may be independently normally distributed in the case of Truckee Meadows (testing was not carried out for Hula Basin because of the possible lack of model fit); and that the standard errors for parameters for both cases were large, indicating nonuniqueness of the solution, even though model fit was good for Truckee Meadows and fair for Hula Basin. The reliability of the predicted drawdown distribution and pumping rate was estimated for a gravel pit operation in Truckee Meadows. In spite of the nonuniqueness problem, predictions of these quantities were not significantly different at a level of significance of 0.05 from the observed values.