The paper develops a structural model and a design methodology for transit system planning in an urban area. Transit "components" are modelled by subarea and by sub-mode in terms of line length, station spacing, and fleet size, in order to determine both quality of service and production costs. Roadway networks are modeled with a Macroscopic Fundamental Diagram that relates speed to network capacity and vehicle demand. Local and global environmental impacts are considered. Travel demand includes both mode-dependent users and mode-choosers able to adopt the mode that offers higher utility. The design methodology involves a mathematical program of welfare optimization with respect to transit factors and fares. Two definitions of welfare are given, one that takes into account only demand surplus and supply profit, the other including environmental impacts. An example of application to Greater Paris shows that there is room for system optimization under current subsidy conditions, and that the explicit inclusion of environmental impacts brings about a significant shift in the "optimal" policy package.