Prediction of areas prone to land degradation in agricultural catchments is a complex task. This is due to the diffi culties encountered in data gathering over wide regions and in the translation of existing scientifi c knowledge to a quantitative and spatially explicit risk assessment system. This paper incorporates the use of remotely sensed data, terrain analysis and a multicriteria mechanism for evaluating risks of soil loss, water ponding, and sediment deposition in a mid-size agricultural Mediterranean catchment, under 80 years of intensive cultivation. The research uses simulations to study the effect of topographic attributes, soil characteristics, vegetation cover, rainfall intensity and human activities on the three above-mentioned processes. The results show that, from the methodological point of view, the integration of knowledge from several experts yields better predictive results than relying on a single expert, even the one found to be most consistent. Also, the use of a simple weighted linear combination was more useful than the more sophisticated computerized programming technique. From the phenomenological point of view, the increase in rainfall intensity and land-use transformation from orchard to fi eld-crops has led to a signifi cant increase in soil loss and sediment yield, while extreme changes in tillage direction have only yielded minor changes in water ponding. The developed system's predictive capabilities also show that the outcomes can be used as a basis for decisions on catchment management in regions of high environmental sensitivity.