In this article, we develop land use models to study the impact of changes in decision variables on soil erosion or other environmental outcomes. From an underlying behavioral model, we use maximum entropy to recover a parametric model of county-level land use shares as a function of decision variables such as output prices, input costs, and land quality. The statistical model may be extended to estimate subcounty land use shares and to incorporate data from federal land use surveys. We use the procedure to analyze the impact of changes in livestock inventories on soil erosion rates in three Iowa counties.