To delay evolution of pest resistance to transgenic cotton that produces Bacillus thuringiensis toxin Cry1Ac (Bt cotton), refuges composed of cotton that does not produce Cry1Ac (non-Bt cotton) are mandated by the United States Environmental Protection Agency. One critical assumption of this strategy is that susceptible adults emerging from non-Bt cotton refuges mate with resistant adults emerging from Bt cotton, which requires movement of adults between Bt cotton and non-Bt cotton. To better understand movements of pink bollworm, the key lepidopteran pest of cotton in Arizona, we analyzed the distribution of wild males and dispersal of sterile males during 1997 on a 259-ha block of a commercial farm containing 76 ha of Bt cotton (69%) and 34 ha of non-Bt cotton (31%). The distribution of 28,397 wild males caught in pheromone traps between 22 May and 20 September suggests that many moved at least 400 m from non-Bt cotton to Bt cotton, yet movement was not sufficient to distribute wild males randomly between Bt cotton and non-Bt cotton. The average number of wild males caught per trap in non-Bt cotton was nearly double that for Bt cotton. Of the estimated 176,000 sterile males released on 8 dates from 18 June through 4 July, we captured 4,447 in pheromone traps between 19 June and 24 July, which represents a recovery rate of 2.5%. The results show that 66–94% of captured sterile males dispersed 400 m or less from the release sites. Along with previously published data, the results reported here suggest that, to promote mating between susceptible and resistant adults, refuges for pink bollworm should be close to Bt cotton.