Wheat plays an important role in the Brazilian commodity production. Therefore, objective and reliable methods for yield estimation are needed particularly at the farm level where several management actions have to be taken. TM Landsat and agrometeorological data were integrated in a model for wheat yield estimation at the farm level for a test site in the south of Sao Paulo State. Landsat data for the crop years of 1986 (three acquisitions) and 1987 (two acquisitions), agronomic and meteorological data were related to yield estimates at the field level (250 fields approximately). Results have shown that the vegetation index derived from TM Landsat explained 60 and 40 per cent of wheat yield variability for the two crop years analysed. The joint use of both vegetation index and agrometeorological data in a single model improved significantly the results as compared to either vegetation index or agrometeorological data separately. The proposed model is to be validated for future crop seasons. Nevertheless, it provided objecctive and reasonably accurate results for wheat estimation on the two crop seasons analysed (R2= 0.65 with a standard error of 339 Kg/ha).