Forest fires are of considerable interest because of their social, economic and environmental consequences. Fire occurrence is influenced by climatic factors: for instance, higher temperatures and atmospheric dryness may increase fire risk. In Central America, because of increasing social awareness, measures have been taken to reduce the occurrence of fires. However, climate change might jeopardize these efforts in the future. For the development of prevention plans, it is important to know how climate change will modify forest fire risk in Central America. Using data mining processes, we investigated the influence of climatic and socio-economic factors on the occurrence of forest fires in Central America. We used climatic monthly data for the 1998-2007 period with a 0.25° spatial resolution and built decision trees to model the occurrence of fires. The decision trees resulted in a good accuracy on the 1998-2007 period. Using climate change and socio-economic scenarios, we applied the same decision trees to future conditions to create maps of future fire risks. Results show that fire risks are decreasing in some areas and increasing in others. The separate and combined effects of socio-economic changes and climate change on the evolution of fire risk are analyzed. The sources of uncertainties are discussed and the approach is evaluated. (Texte intégral)