The occurrence and intensity of an epidemic are determined by the interactions of the host with the pathogen or the pest, the environment and the agronomic management (as shade management in coffee agroforestry systems). As a consequence of the spatial heterogeneity of these factors, patchiness is the rule in the distribution of plant pests and diseases. Environmental information is now readily available in high resolution and can be combined with spatial analyses to determine potential pest and disease distribution due to environmental factors, and subsequently lead to better decisions and improved risk management. The objective of this paper is to show how better decisions and disease riskadapted agroforestry practices, for entire coffee growing regions, can be derived based on spatial decision support tools and a minimum of ground data evidence. We used ground data, on coffee pests and diseases, collected in previous surveys conducted in Costa Rican coffee plots within a range of shade density. The diseases retained for our analyses were coffee rust (Hemileia vastatrix), American leaf spot disease (Mycena citricolor), and coffee blight (Phoma costarricencis). We first used the environmental data for the coffee plots with less shade density, and generated for the different diseases the decisive environmental driving factors by means of Geographical Information System (GIS). The climatic data such as radiation, precipitation and temperature are derived on a 1 km resolution. We used algorithms based on maximum entropy, Bayesian statistics, and spatial analysis to delimit areas with distinct risk potential. The results appraise the disease risk of coffee growing areas associated with their physical characteristics. For the areas where the results were significant, the decisive factors for each disease are identified and shade-management strategies are suggested according to their known effect on these factors. (Texte intégral)