Annual bromes (downy brome and Japanese brome) have been shown to decrease perennial grass forage production and alter ecosystem functions in northern Great Plains rangelands. Large-scale chemical control might be a method for increasing rangeland forage production. Although fall application has been shown to be the most effective and least likely to impact co-occurring native species, spring germination of downy brome may reduce the efficacy of fall-only herbicide application. We assessed the impact of a low glyphosate dose rate (210 g ha−1) applied to rangelands in fall or in fall and spring on nontarget species and on annual brome abundance at two sites in eastern Montana over 2 yr. We tested the following hypotheses: (1) nontarget effects are greater with spring herbicide application, (2) fall and spring herbicide application are necessary for effective downy brome control, and (3) fall herbicide application is sufficient to control Japanese brome. Few nontarget effects occurred; two dicotyledonous species exhibited small increases in response to herbicide. We found that that a single fall application reduced downy brome cover and seed bank density, but after the second fall application in the following year, downy brome did not continue to show a response to herbicide. After 2 yr of fall herbicide application, Japanese brome had denser seed banks in plots where herbicide had been applied. Blanket glyphosate application on rangelands is an unreliable method for controlling annual brome invasions in the northern Great Plains.