This paper describes a methodology for providing systematic land use/land cover classification on a regional scale, with specific application to the High Taquari Basin, in the state of Mato Grosso do Sul, Brazil. The methodology is based on previous work developed at INRIA by Bouzidi et al. (2000) and proposes the use of multitemporal 250m resolution surface reflectance data from the Moderate Resolution Imaging Spectroradiometer (MODIS), on board NASA's Terra satellite. This data will be coupled with higher-resolution training information in order to compute characteristic temporal MODIS reflectance profiles for each class and each band. A linear mixture model will then be applied so as to compute the proportions of each class within each MODIS pixel over areas where classification information may not be available, thus enabling the monitoring of land use/land cover over large regions such as the High Taquari Basin and its adjacencies.