This dissertation considers issues related to the traditional urban travel demand forecasting procedure and the development of combined network equilibrium models of urban residential location and travel choices. Three hierarchical models combining the traditional trip distribution, modal split and traffic assignment models are formulated. In these models, dispersion measures related to conditional travel choices are used as constraints in the model framework; the resulting conditional choice probabilities, which are related in a nested way, reflect hypothesized decision processes in reality. The parameters of these models are estimated by the full information maximum likelihood method based on endogenously-determined user-equilibrium travel costs (auto travel times and operating costs) since no data for these costs exist. The estimation procedure is constructed as a bi-level programming problem in which the upper problem is to estimate the parameters using the maximum likelihood method and the lower problem solves for the choice probabilities and costs given the parameter values. The bi-level programming problem is solved iteratively to obtain the best goodness-of-fit of the model's prediction to observed data. All parameters, including dispersion parameters, generalized cost coefficients and balancing factors in the doubly constrained gravity model, are estimated simultaneously on a sketch planning network of the Chicago region with 317 zones, 1060 nodes and 2902 highway links. The resulting models are compared in terms of their relative performance on base year data.