All herbicide treatments controlled the initial stand of Canada thistle [Cirsium arvense (L.) Scop.] and musk thistle (Carduus nutans L.) on a subirrigated range site. Annual retreatment controlled seedlings and the few plants that emerged from surviving root remnants after the first application. Average perennial grass production on unfertilized, herbicide-treated plots increased 110, 314, and 212%/yr over unfertilized check plots during the 3-yr period of treatment, i.e., 960, 3450, and 4300 kg/ha, respectively. Grasses did not fully reoccupy the site at the end of 3 consecutive yr of excellent thistle and forb control. The increase in grass production varied considerably among several herbicide treatments that controlled thistles equally. Grass production in the third year was greater under the 3,6-dichloropicolinic acid and 2,4-D [(2,4-dichlorophenoxy)acetic acid] treatment series than all other treatments. Ammonium nitrate enhanced the recovery of grass damaged by certain herbicides, but favored the competitive ability of the thistle more than that of the grasses.