The anthropogenic deforestation occurring throughout the tropical world is one of the great crises of our time. To better understand the dimensions of this problem, a number of studies have sought to quantify the extent and rate of tropical forest loss. However, environmental planners and managers need to know more than the extent and rate of tropical deforestation. There is a need for predictability, i.e. which areas are most susceptible to deforestation. This study demonstrates a methodology for predicting those areas with the greatest propensity for deforestation based on natural and cultural landscape variables. Logistic regression analysis was used to determine variables most closely associated with deforestation. The independent variables most strongly associated with the forested/deforested dependent variable were location of forested areas in relation to the forested/deforested edge in an earlier time period, to the location of access in an earlier time period, and to the location of habitation in an earlier time period. GIS analysis was then used to verify spatially the close statistical relationship between the dependent variable and each of the independent variables selected by the logistic regression modeling.