Savannas cover one fifth of the earth’s surface and is home to one billion of the world’s population. It is therefore no surprise that they are experiencing a wealth of land use pressures resulting from population growth. Landsat is one tool used to map land use change in savannas. While those will argue Landsat does not offer the high spatial resolution of other forms of remote sensing, it does provide the necessary data dating back to the 1970s which is valuable for multi-temporal analysis. Maximum Likelihood image classification (ML), NDVI, change detection and field validation were all used in the classification of Landsat subsets and Landsat country mosaics of Belize between 1980 and 2010. The classified 2010 Landsat mosaic accuracy was assessed using a confusion matrix. The results of the research confirmed the capabilities of Landsat imagery for mapping savannas and their land use. The classification of forest and savanna along with major land use pressures from agriculture and aquaculture were properly identified. Many attempts prevented the identification of sub agricultural crop types, urban areas, and micro savanna habitats. The results conclude an underestimation of the savannas using ML classification on Landsat data for both the 1980 and 2010 data sets.