Quantifying the impact of land use on water supply and quality is a primary focus of environmental management. In this work we apply a semidistributed hydrological model (SWAT) to predict the impact of different land management practices on water and agricultural chemical yield over a long period of time for a study site situated in the Arborea region of central Sardinia, Italy. The physical processes associated with water movement, crop growth, and nutrient cycling are directly modeled by SWAT. The model simulations are used to identify indicators that reflect critical processes related to the integrity and sustainability of the ecosystem. Specifically we focus on stream quality and quantity indicators associated with anthropogenic and natural sources of pollution. A multicriteria decision support system is then used to develop the analysis matrix where water quality and quantity indicators for the rivers, lagoons, and soil are combined with socio-economic variables. The DSS is used to assess four options involving alternative watersheds designated for intensive agriculture and dairy farming and the use or not of treated wastewater for irrigation. Our analysis suggests that of the four options, the most widely acceptable consists in the transfer of intensive agricultural practices to the larger watershed, which is less vulnerable, in tandem with wastewater reuse, which rates highly due to water scarcity in this region of the Mediterranean. More generally, the work demonstrates how both qualitative and quantitative methods and information can assist decision making in complex settings.