The efficacy of glyphosate (Roundup-Ultra® herbicide) for the control of purple and yellow nutsedge (Cyperus rotundus and C. esculentus, respectively) in Roundup Ready® cotton was evaluated in field studies conducted at the University of Arizona Maricopa Agricultural Center in 1995 and 1996. The experimental design used for the nutsedge studies was a randomized complete block design with five (1995) or 6 (1996) blocks in which individual plots were four 40 inch rows by 40 feet long. The soil at the field study site was a Trix clay loam that contained 1.4 percent organic matter, 42 percent sand, 24 percent silt, and 34 percent clay. Prowl (pendimethalin) or a tank-mix of Prowl and Zorial (norflurazon) were applied to flat ground and incorporated with a spring tooth harrow consisting of three ranks of spring steel teeth a few weeks before planting (see data tables for rates in lb of active ingredient per acre). The fields were listed and the large peaked beds were rotomulched. The seed beds were mulched a second time using a Sidewinder rotomulcher and then shaped with a bed shaper. Roundup Ready cotton seed was planted about 0.5 inches deep in dry soil on April 18, 1995 and April 25, 1996 and then were irrigated to initiate germination. Postemergence herbicide treatments were applied to the plots with tractor mounted equipment, either a gasoline motor driven roller pump sprayer, or a compressed air plot sprayer. See data table for herbicide treatment rates (lb a.i./A), application dates, cotton growth stage at the time of each application, and the dates weed control efficacy data was collected. In 1995, a special formulation of Roundup herbicide (MON 2139) that contained glyphosate without surfactant was used and 0.05 percent (v/v) of a non-ionic surfactant (MON 0818) was separately added to the spray solutions. In 1996, Roundup-Ultra was used without additional surfactants. Standard University of Arizona cotton production and insect control recommendations were followed to produce the cotton crop. The center two rows of the four row plots were machine harvested on October 17, 1995 and November 1, 1996. Analysis of variance and mean separation with Duncan's multiple range test were used to elucidate differences between treatments.