The authors use a dynamic stochastic simulation model of forage, herbivores, predators and domestic livestock in the Machakos District of Kenya to address policies related to the multiple use of rangeland resources. The particular policy examined is that of switching from a traditional system, where commercial ranchers do not harvest wildlife herbivores, to one where ranchers are provided economic incentives to adopt multiple-use management of the range resource. Simulations using an adaptive controller indicate that the effects of the policy change on wildlife populations depend on the conditions of the ecosystem and, importantly, on ranchers' attitudes to risk. When forage is abundant, and game and livestock do not compete for food, the policy change leads to reduced wildlife populations, especially of the relatively more valuable species. However, in periods of drought when competition for forage occurs, the policy change may dampen the decline in game populations, as risk-averse ranchers may decide to sell more cattle and harvest wildlife instead. Game cropping reduces wildlife populations, but increases their stability.