Biologists now engineer transgenic crop varieties that express proteins that are toxic to a variety of common agricultural pests. These transgenic crops offer farmers a new tool for effectively managing pests that reduce yields and increase production costs. However, the concern over pest resistance to these toxins has prompted the EPA to require resistance management plans. Seed companies have focused on a high-does refuge plan where farmers are required to plant a constant proportion of cropland in refuge in order to maintain a susceptible pest population. Currently, entomologists recommend 20 to 40% refuge. This paper develops an economic model of pest management with pest resistance to estimate the constant proportion of refuge that maximizes farm income over a fixed planning horizon. Results indicate that there is a clear economic tradeoff between the pest control and population management benefits afforded by a transgenic variety and the resistance management benefits and savings in production costs afforded by refuge. Under certain circumstance a 20 to 40% refuge is economically sensible. However, the optimal proportion of refuge is sensitive to a number of uncertain biological factors: the initial frequency of resistant pests, and the survival rate of resistant and susceptible pests. Additionally, we find that when the pest population and resistance develop slowly, the economic losses of a suboptimal proportion of refuge are relatively small; however, the biological consequences in terms of pest susceptibility are very large.