A 3-yr dryland study was initiated in 1999 at Munday, TX, on an Altus fine sandy loam to determine the most appropriate cotton growth stage or stages at which to apply glyphosate. The objectives were: (1) to evaluate control of silverleaf nightshade in the cotton and (2) to determine the effect on yield. Treatments were: (1) control (C; two cultivations), (2) early glyphosate at the four-leaf stage (E), (3) early glyphosate followed by a midseason application 21 d later (E+M), (4) two cultivations plus late glyphosate at 20% open bolls (L), (5) treatment E+L, and (6) treatment E+M+L. In fall 1999 and 2001, silverleaf nightshade stem numbers decreased in the plots sprayed early and midseason, and increased in plots receiving only late or no applications when compared with the counts in the spring of those years. However, in fall 2000, nightshade numbers were less than in early spring 2000 regardless of treatment, probably because of hot and dry weather. In fall 2001, silverleaf nightshade populations had increased 13-fold and twofold for the C and L treatments, respectively, when compared with populations at the beginning of the study. Other treatments had population decreases of 10 to 90%. Three-year average lint yields were higher with early or early plus midseason applications. Lint yields were similar with early or early plus midseason sprays. Lint yields were higher when nightshade/cotton biomass competition was lower. Early application of glyphosate can effectively control silverleaf nightshade populations and can increase yield when compared to no application or a late application.