Previous studies have suggested that market failures are household-specific and not commodity-specific (de Janvry et al, 1991); transaction costs determine whether a household is a buyer, seller or self-sufficient for a given good and how much it is going to produce (Key et al, 2000). Focusing on fuelwood production in northern Uganda, this paper extends previous studies by introducing fixed transaction costs associated with reaching the market and the forest. We predict that households sort in space, with autarkic households being located closest to the forest and farthest from the market, buyer households located closest to the market and farthest from the forest and seller households located at intermediate distances from the market and forest. We show that the spatial predictions hold in partial and general equilibrium settings. We test the predictions of our model using data from northern Uganda and find evidence that supports the predictions from our theoretical model. The ensuing spatial-dynamic simulations based on the static allow us to make forecasts of where forest degradation is likely to occur as well as to model spillover effects resulting from the introduction of a conservation intervention like a protected area.