Parking policy is still the most directly available instrument for managing traffic demand in many cities. But policy design is subject to difficulties resulting from the complexity of the urban mobility system. This article presents a model framework, based on a system dynamics approach, aimed at assessing the effectiveness of parking policy and quantitatively identifying optimal design at an aggregate spatial level under a level of service maximization objective. An application to a city is developed and the results are discussed in view of their qualitative outcomes and quantitative validity and robustness. It is argued that system dynamics addresses several needs of modellers and decision makers regarding urban parking policy assessment, particularly if parking is used as a traffic management tool. The qualitative results of the model coincide with the prescriptions that would come from the economic theory, even with an objective function based on level of service instead of a broader indicator of efficiency. At the quantitative level, the validation testing of the model application with the available data provided positive indications and no case to reject that a quantitative accurateness useful for policy prescription could be attained provided that some data gaps are fulfilled. The necessary data for calibration seems to be possible to obtain by feasible local empirical observations.