In rangeland ecosystems, invasive annual grass replacement of native perennials is associated with higher fire risk. Large bunchgrasses are often seeded to reduce cover of annuals such as Bromus tectorum L. (cheatgrass), but there is limited information about how revegetation reduces fire risk over the long term. We assessed how revegetated community composition influences fire risk at three sites in Columbia National Wildlife Refuge in Grant County, Washington that were revegetated with large bunchgrasses 8 - 18 years before the study. At each site, five replicates of 10 plots (10 × 10 m) were established. Fire risk was determined as the probability that a plot would completely burn following ignition at a randomly located point in each plot (i.e., if 8 of 10 plots burned, then fire risk was 80%). Preignition, cover of bunchgrasses, cheatgrass, forbs, and surface characteristics were determined for each plot. Fire risk was < 100%. However, fire risk was still relatively high around 73% and did not differ significantly among sites despite differences in cheatgrass and bunchgrass cover, which may have been attributable to other characteristics, such as high total fuels cover (> 80% at all sites) and unvegetated gap cover (soil and soil cryptogams, < 17%). This information can provide guidance for future studies with larger ranges of cover characteristics to develop robust fire risk models, which ultimately will be used to aid rangeland managers who need to specify reduction of fire risk after reestablishing large bunchgrasses in rangelands infested with cheatgrass.