In Central Aragon, winter cereal is sown in the autumn (November December), commonly after a 16-18 months fallow period aimed at conserving soil water. This paper uses the Simple SoilPlantAtmosphere Transfer (SiSPAT) model, in conjunction with field data, to study the effect of long fallowing on the soil water balance under three tillage management systems (conventional tillage, CT; reduced tillage, RT; and no-tillage, NT). This was on the assumption that soil properties would remain unchanged during the entire fallow season. Once the model was validated with data obtained before primary tillage implementation, the differences between simulated and observed soilwater losses for the CT and RT treatments could be interpreted as the direct effect of the soil tillage system. The model was calibrated and validated in a long-term tillage experiment using data fromthree contrasting long-fallowseasons over the period 1999-2002, where special attention was paid to predicting soil hydraulic properties in the pre-tillage conditions. The capacity of the model to simulate the soil water balance and its components over long fallowing was demonstrated. Both the fallow rainfall pattern and the tillage management system affected the soil water budget and components predicted by the model. The model predicted that about 81% of fallow seasonal rainfall is lost by evaporation in long-fallow periods with both a dry autumnin the first year of fallowand a rainfall above normal in spring. Whereas, when the fallow season is characterised by a wet autumn during the first year of fallow the model predicted a decrease in soil water evaporation and an increase in water storage and deep drainage components. In this case, the predicted water lost by evaporation was higher under NT (64%) than under RT (56%) and CT (44%). The comparison between measured and simulated soilwater loss showed that the practice of tillage decreased soil water conservation in the short term. The longterm analysis of the soil water balance showed that, in fallow periods with a wet autumn during the first year of fallow, the soil water loss measured under CT and RT was moderately greater than that predicted by the model.