Degradation of arid rangeland, and efforts to control that degradation, have become topical issues. However, the inherent characteristics of the rangeland, and the intertemporal nature of the problem, complicate the analysis of degradation issues in the search for more appropriate rangeland policies. Stochastic dynamic programming is examined as one means of allowing for those complexities. Using the case of the Queensland mulga rangelands, optimal stocking rates are shown to rise with lower property sizes, higher discount rates, higher wool prices and declining risk aversion. Importantly, the analysis reveals that a strategy of high stocking rates with the potential for rangeland degradation is an optimal response to the economic and social factors that confront graziers and is not an intertemporal information problem alone.