Erosion and sediments are among the most important externalities in the developing world. These sediments negatively affect the quantity and quality of water in the downstream regions of watersheds. In light with the growing interest in many developing countries to use market-based instruments, this paper develops a model for designing efficient environmental policy at a watershed scale. Because farm households are heterogeneous in a given watershed, we develop a spatially explicit, heterogeneous watershed scale environmental policy to lesson watershed degradation. We use GIS data and geo-referenced household plots to populate the watershed with the heterogeneous households. Heterogeneity also implies that the impact of environmental tax policy on poverty varies among households. The model results confirm the benefits of differentiating policy based on the spatial attributes of the watershed. Our study proposes the possibility of funding poverty reduction using the revenues from environmental taxes. The results show that, for a moderate reduction in soil erosion, revenues from environmental taxes could be used for poverty reduction. However, for larger improvement in environmental quality, the efficient environmental tax would not be sufficient to compensate the poor. Our findings reveal the extent of tradeoffs between poverty reduction and environmental protection. In other words, tighter environmental policies could exacerbate poverty unless assistance is provided to the poor.