Stink bugs (Hemiptera: Pentatomidae) were sampled in commercial cotton, Gossypium hirsutum L., fields in 2007 and 2008 in South Carolina and Georgia. The main species collected with the beat cloth and sweep net methods were green stink bug, Acrosternum hilare (Say) (63 and 57%, respectively); brown stink bug, Euschistus servus (Say) (23 and 18%, respectively); and southern green stink bug, Nezara viridula (L.) (14 and 22%, respectively). Average stink bug densities were 0.145 ± 0.010 (mean ± SEM) for adults and 0.250 ± 0.027 for nymphs per 3.7 m of row by using the beat cloth method. Average stink bug densities were 0.291 = 0.016 for all adults and 0.137 ± 0.018 for all nymphs per 50 sweeps. A density of two southern green stink bugs per 3.7 m of row by using the beat cloth method required 43 samples (reliability or precision, D_x = 0.3) for population estimation, whereas 88 samples were necessary for a density of two southern green stink bugs per 50 sweeps. At low densities, the sweep net was a more cost-reliable sampling method for all species and life stages. For adult stink bugs, the beat cloth method became more reliable at densities of 1.0, 3.2, and 5.8 stink bugs per 3.7 m of cotton row for southern green stink bug, brown stink bug, and green stink bug, respectively. Sequential sampling consistently reduced sample size for all insects compared with a fixed sampling plan.