The tarnished plant bug, Lygus lineolaris (Palisot de Beauvois), is an important pest of cotton, Gossypium hirsutum L., in the mid-southern United States. It is exclusively controlled with foliar insecticide applications, and sampling methods and thresholds need to be revisited. The current experiment was designed to establish a plant-based threshold during the flowering period of cotton development. Experiments were conducted in Mississippi in 2005 and 2006, Arkansas in 2005, and Louisiana in 2005 through 2008. Treatments consisted of various combinations of thresholds based on the percentage of dirty squares that were compared with the current threshold with a drop cloth or automatic weekly applications. Dirty squares were characterized as those with yellow staining on the developing bud resulting from tarnished plant bug excrement. Treatments consisted of 5, 10, 20, and 30% dirty squares. Each plot was sampled weekly, and insecticides were applied when the mean of all replications of a particular treatment reached the designated threshold. At the end of the season, plots were harvested and lint yields were recorded. Differences were observed in the number of applications and yields among the different treatments. The 10% dirty squares threshold resulted in a similar economic return compared with the drop cloth. A threshold of 10% dirty squares resulted in a similar number of insecticide applications, yields, and economic returns compared with that observed with the drop cloth. These results suggest that a threshold of 10% dirty squares could be used to trigger insecticide applications targeting tarnished plant bugs in flowering cotton.