The objective of this work was to monitor and estimate soybean grain yield based on an agronomical model coupled to a Geographic Information System (GIS) using meteorological data generated by the Eta mesoscale weather forecast model. The agronomic model produced bi-weekly penalizing indices maps and soybean grain yield estimates for the States of Mato Grosso do Sul, Paraná and São Paulo, Brazil, during crop year 2004/05. Result showed that meteorological data from the Eta model, incorporated in an agronomic model, allowed monitoring, in time and space, the soybean crop. In addition, the meteorological data from Eta model are provided in almost real time allowing a quantitative and quick evaluation of weather conditions for soybean production.