The spread of non-indigenous, C4 grasses threatens global conservation of savannas and subtropical grasslands. Identifying control methods to selectively target these invasives has proven difficult. Here, we tested the hypothesis that the effectiveness of prescribed burns for control is determined, in part, by the phenology of the target species at burn time. We conducted two experiments in a subhumid, C4 grassland in central Texas. The focal invasive was the C4, perennial bunchgrass Bothriochloa ischaemum (L.) Keng (KR bluestem). Burns were conducted in early and late fall when plants were in different phenological states. In addition, we attempted to manipulate phenological state through temporary rainout shelters to expedite maturation. The two experiments differed in the timing of the rainout shelter application (experiment 1: May to July, experiment 2: August and September), but otherwise had the same complete factorial design (burn time × shelter). Across experiments, when at least 50% of all tillers were pre-reproductive at burn time, either due to shelter treatment or time of year, spring tiller densities were significantly lower than when plants were burned in a more advanced reproductive state. Trends in fall biomass generally followed trends in spring tiller densities, with one exception where plants in no-shelter plots burned in October had lower biomass than expected based on tiller densities. Treatment responses for the native C4 grass B. laguroides were consistent with those of B. ischaemum. These findings suggest that strategic burns can be used to reduce the subsequent recovery of invasive C4 grasses while not disadvantaging native grasses.