A methodology is described to realize an Intelligent Agent as a planning tool to provide information on man-induced environmental impacts examining the effects on groundwater quality possibly deteriorated by chemical, physical, and biological pollution from civil, industrial and agricultural activities. The planning domain often has high information incompleteness and the use of complex mathematical models to simulate the phenomena in a territorial system becomes very difficult. To overcome such limits, the outcomes of the models have to be integrated with more qualitative representation, interacting with experts capable of resolving a complex problem using their 'expertise'. A pollution risk evaluation methodology was developed that combines the value of intrinsic vulnerability of a local aquifer, through mathematical equations of standard procedures (CNR - GNDCI method) and a parametric managerial model (SI NTACS), and the human activities-related hazard value using the experts' knowledge.