Increases of velvet mesquite (Prosopis velutina Woot.) in southwestern grasslands might have been caused by livestock consumption of fuels that once burned with sufficient frequency and intensity to kill the trees. However, attempts to control mesquite with fire usually have failed. We measured fire damage and 5 years of postfire recovery for 225 mesquite trees > 1 m tall, following a 2002 wildfire that included grasslands differing in fire history, presence vs. 34-year livestock exclusion, and predominance of native vs. exotic grasses. The fire burned 100% of ground cover in ungrazed areas and 65% on grazed lands. Top-kill was 100% for trees in exotic ungrazed grasslands (the areas with highest fuel loads), 79% for trees in ungrazed native grasslands, and 28% for trees in grazed grasslands. Most top-killed trees produced ground sprouts, so that by 2006 the combined foliage volume from ground sprouts and surviving branches was 78% (± 3.2 SE) of preburn foliage volume in grazed areas, 66% (± 3.3) in ungrazed exotic grasslands, and 57% (± 4.0) in ungrazed native grasslands. Fire damage was greater among surviving trees in ungrazed areas that had burned twice (1987 and 2002) than among those that had burned only once since 1968 (in 2002), especially in native grasslands where postfire foliage recovery for twice-burned trees was only 47% (± 6.3) by 2006. Only 1 of 84 trees died in the area burned once, whereas 12 of 66 (18.2%) died in the area burned twice, including several individuals > 3 m tall. These results suggest that repeated fires likely could have prevented the historic spread of velvet mesquite into southwestern grasslands, but probably could be used to control mesquite today only in areas where abundant herbaceous growth provides sufficient fine fuels.