Kentucky bluegrass (Poa pratensis L.) invades northern Great Plains rangelands. On the Sheyenne National Grassland in southeastern North Dakota, three research sites, each with a different level of Kentucky bluegrass invasion, were chosen to evaluate effectiveness of burning and burning–herbicide combinations to control Kentucky bluegrass. Initial Kentucky bluegrass invasion levels were 37%, 77%, and 91% for LOW, MODERATE, and HIGH invaded sites, respectively. Within each invaded site, four replicated strips (20 by 60 m) were established, with half of each strip burned in late October 2005 and the other half burned in early May 2006. Herbicide treatments of (1) no herbicide, (2) 2.24 kg ha−1 of glyphosate, and (3) 0.43 kg ha−1 of imazapic were randomly assigned to 10 by 20 m subplots within each burn. Control plots were established at the same time. Relative basal cover of native grass, native forb, and Kentucky bluegrass was estimated annually using 50 10-point frames within each subplot. On the HIGH site in 2006, fall-burned plots with a spring glyphosate application had three times the native grass cover and only one fourth of the Kentucky bluegrass cover compared with controls. Similar results with the same treatment occurred at the MODERATE site. Native grasses became the most abundant plant community on these plots in the MODERATE and HIGH sites within 1 yr. Treatment differences were transitory, and the LOW site differed from the MODERATE and HIGH sites. In 2007, on fall-burned plots with spring glyphosate application, the amount of Kentucky bluegrass was 14% and 30%, and native grass species were 52% and 42% on the MODERATE and HIGH sites, respectively, which was similar to the initial values on the LOW site. These data emphasize the importance of initial invasion level in developing restoration strategies and provide evidence burning and herbicide combinations can be valuable management tools even on heavily invaded grasslands.