In Honduras, traditional coffee processing is the cause of two problems: poor coffee quality and contaminated water. In this study we propose to replace traditional coffee processing plants with a network of improved ecological plants that would be optimally located in a sub-watershed. The method is an adaptation of a spatial integer linear programming that determines the optimal location and size of new coffee processing plants. We applied the method to a typical sub-watershed in the hillsides of western Honduras and show that coffee quality can be improved and contamination can be reduced substantially at a relatively low cost. We also calculated the incentive for small farmers to give up home processing. We find that the incentive is much lower than the premium that could be obtained from an improved coffee quality.