This study examined the efficient refuge policies for Bt cotton for three cotton growing regions in India. This was accomplished by developing a single-pest, dual-toxin biological model simulating bollworm resistance to the Bt toxin and synthetic pyrethroids, followed by formulating profit functions for Bt and non-Bt cotton for a representative producer in each region. Profits received in subsequent periods were considered in the regulatory model in order to choose a refuge constraint (static problem) or a sequence of refuge policies (dynamic problem) for each region that maximize discounted profits received over 15 years, subject to various economic and biological constraints. Dynamic solutions for the regulatory problem were derived for each region using the Bellman equation. Results suggested that South Indian farmers do not need to grow a refuge, but farmers in the North and Central regions do. Results also suggested that planting sprayed refugia might be more profitable than planting unsprayed refugia. Sensitivity analysis revealed that the refuge requirements were sensitive to the initial Bt resistance level, relative proportion of CBWs in natural refuges, and proportions of heterozygous and homozygous fitnesses in all of the three regions. Moreover, static refugia were found more profitable as compared to dynamic refugia in the North and Central regions.