Firstly, the coordination between land-use and urban transport system is reviewed and discussed in this paper. It is the private transport system and the public transport system that constitute urban transport system. After analyzing the difference between the private transport and the public transport oriented development mode, we make a conclusion that a optimized transport structure can be acquired if we get a dynamic equilibrium. Then in the view of the issue of the urban development and the demand of private transport and the public transport, a case study is conducted.This study established the forecast model of urban car ownership in terms of transportation demand and environment capacity, analyzed the economy effect of road traffic based on economy and diseconomy aspects induced by car industry and car use respectively, to provide theory and practical base for sustainable transportation system. In this model, a macro-level forecast model of urban car ownership based on artificial neural network technique is developed in the first. To simulate the non-linear relationship between car ownership and its effect factors, the main factors are analyzed, and seven input vectors, such as per-capita GDP, transit network density, road network density, car price/income, patrol price and external disturbing factors, are taken as independent variables and car ownership is taken as output variable to establish the forecast model based on BP neural network technique. Here, external variable, including some affecting factors which could not be measured, is input to the model to simulate the sudden effect induced by them.Secondly, taking traffic environment load into account, the maximum urban car ownership model constrained by transportation environment capacity is developed. The purpose is to optimize the traffic distribution and get the maximum car ownership constrained by environment capacity. A bi-level programming problem has been established. The upper level is a car ownership model, in which objective function is to maximize the total of zonal car ownerships and the constraint is traffic environment load on a link should not exceed the pre-decided capacity. The lower level is a fixed demand user equilibrium assignment model, which simulates the travelers' path choice behavior and calculates distributions and running characteristics of traffic demand on road network. To realize the feedback between the two levels and solve the optimization problems simultaneously, we developed an optimal algorithm based on sensitivity analysis, namely acquire derivative function of link volume and traffic demand with respect to zonal car ownership, and feedback the function into upper level program.