We develop an artificial ecology that simulates the interaction between hunter decisions and prey behaviour, using ibex hunting in the North Tien Shan Mountains as a case study. The aim is to model hunter and prey behaviour at a low enough level that overall population dynamics and hunter costs are emergent properties of the system rather than being assumed, as is usually the case. A genetic algorithm linked to artificial neural networks is used to evolve hunter decisions about where to hunt. We demonstrate the importance of the number of people hunting, which is determined by the profitability of hunting, as a key driver of system dynamics. A fundamental difference emerges between outcomes on approach to equilibrium, and after stochastic equilibrium had been reached, with extinction being common on approach and virtually non-existent thereafter. This probably reflects naive ibex behaviour on commencement of hunting. The framework developed here is flexible and transferable, and is particularly useful for the strategic testing of management strategies.