The success of an automatic calibration procedure is highly dependent on the choice of the objective function and the nature (quantity and quality) of the data used. The objective function should be selected on the basis of the stochastic properties of the errors present in the data and in the model. Also, the data should be chosen so as to contain as much valuable information about the process as possible. In this paper we compare the performance of two maximum likelihood estimators, the AMLE, which assumes the presence of first lag autocorrelated homogeneous variance errors, and the HMLE, which assumes the presence of uncorrelated inhomogeneous variance errors, to the commonly used simple least squares criterion, SLS. The model calibrated was the soil moisture accounting model of the U.S. National Weather Service's river forecast system (SMA-NWSRFS). The results indicate that a properly chosen objective function can enhance the possibility of obtaining unique and conceptually realistic parameter estimates. Furthermore, the sensitivity of the estimation results to various characteristics of the calibration data, such as hydrologic variability and length, are substantially reduced.