Agroforestry can help prevent land degradation while allowing continuing use of land to produce crops and livestock. A problem with the evaluation of agroforestry using long-run static models and traditional discounting techniques is that the present value of the forestry enterprise is generally much lower than that of other production activities. This problem is common with Australian native species which tend to have a high environmental value but a low market value. This paper presents an economic analysis of an agroforestry operation in land prone to degradation and in the presence of positive externalities provided by trees. The value of the land is estimated based on the present value of expected returns in perpetuity under optimal management. Simulation analysis is used to evaluate the loss in land value caused by dryland salinity. A nonlinear programming model is developed and used to study the effects of timber prices and forest planting costs on optimal forest area and the level of salinity. Elasticities of relevant variables with respect to prices and costs are derived and policy implications of results are discussed.