The reliable prediction of the outbreak of epidemics is of great importance to the health services of the developing countries. In the case of sub-Saharan Africa conventional routine malaria control programs are complex and unaffordable. The prediction of an outbreak a few months ahead will allow intervention by vector control and by planning large-scale chemotherapeutic treatment. The development of new tools for medium-range weather forecasting offers the possibility that epidemics be predicted many months in advance (even prior to the start of the rainy season). The European Union DEMETER project, sets out to test the skill and reliability of multi-model seasonal forecasts which are based on an ensemble of general circulation models. Using a weather driven mathematical model of Anopheles mosquito and Plasmodium falciparum parasite population we intend to test the value of these probabilistic forecasts to the malaria control community in terms of improved temporal and spatial prediction of malaria epidemics. Here we present the modeling structure and the preliminary results.