This article examines the implications of upscaling tropical deforestation for climate change. In this case, upscaling refers to the extrapolation and aggregation of deforestation to the grid scale that is used in global climate models (GCMs). The upscaling of deforestation emphasizes the extent of forest loss, and assumes that deforestation is a homogeneous and instantaneous process. The structure of deforested landscapes is usually disregarded in "upscaled" experiments, and the intensity of deforestation is seldom considered. Consequently, the atmospheric response to a heterogeneous surface is not addressed. Furthermore, climatically significant soil and vegetation parameters associated with complex and dynamic deforested landscapes are ignored. These factors underscore the need for more realistic representation of tropical deforestation in modeling studies. Several recent attempts to address the issue of scale in deforestation studies are described in the article.