The agricultural expansion in Mato Grosso state, Brazil, is a process that in recent decades has been consolidated due to the high production of grains such as soybeans. This process is closely related to the land use and land cover changes and the replacement of natural forest cover for economic activities such as livestock and agriculture, especially, along the southern and eastern edges of the Amazon region. In this sense, several scientific projects and researches aim to support important decision-making using remote sensing data and techniques to study the land use and land cover changes. The change detection techniques allow the recognition of agricultural expansion patterns and it represents an interesting tool to improve territory management and public policies. This present study aimed to analyze the spatial-temporal trajectory of agricultural expansion patterns in an Amazon region applying a change detection technique and using the land cover classifications generated by TerraClass in different years (2004, 2010, 2014). The classification results were considered satisfactory (kappa greater than 0.7 and overall accuracy higher than 0.8) and indicate further consolidation of agricultural expansion process, as well as continued expansion of annual agricultural patches, which leads to a better assessment and discussion about Amazon cover changes.