Abstract Environmental Agencies require Decision Support Systems, in order to plan Air Quality Policies considering the cost of emission reduction measures and the human health effects (with related social costs). The use of Decision Support Systems is also useful to spread information to general public, explaining the effectiveness of proposed air quality plans. In this paper, a multi-objective approach to control PM10 concentration at a regional level is presented. The problem considers both the internal costs (due to the implementation of emission reduction measures) and the external costs (due to population exposure to high PM10 concentrations). To model PM10 concentrations, a single surrogate model is used for the entire domain, allowing the implementation of a very efficient optimization procedure. The surrogate model is derived through a set of 10 simulations, performed using a Chemistry Transport Model fed with different emission reduction scenarios. The methodology is applied to Northern Italy, a region affected by very high PM10 concentrations that exceed the limit values specified by the EU legislation.