Field experiments were conducted in 2003, 2004, 2005, and 2006 to determine the critical timing of Palmer amaranth (Amaranthus palmeri S. Wats.) removal in second-generation glyphosate-resistant cotton (Gossypium hirsutum L.). Treatments consisted of nine durations of Palmer amaranth interference that ranged from 0 to 63 d after emergence (DAE) in increments of 7 d. Glyphosate was applied to remove Palmer amaranth at the appropriate timing for each treatment, and plots were maintained weed-free for the remainder of the season. Cotton lint yield loss increased gradually (from 0 to 3%) when Palmer amaranth removal was delayed from 0 to 21 DAE and increased rapidly (from 3 to 77%) when removal was delayed from 21 DAE to 63 DAE. Palmer amaranth biomass increased slightly when timing of removal was delayed from 0 to 21 DAE and increased dramatically when removal was delayed beyond 21 DAE. Palmer amaranth biomass weights were 0, 96, and 1810 g m-2 when removal occurred 0, 21, and 63 DAE, respectively; furthermore, regression of cotton lint yield loss as a function of Palmer amaranth biomass revealed that the two variables were strongly correlated. Using a cotton lint yield loss threshold of 2.7%, which was calculated using cotton lint yield, cotton lint price, and cost of glyphosate, the critical timing of Palmer amaranth removal in second-generation glyphosate-resistant cotton was 19 DAE.