Mitigating invasive annual grass impacts is critical to halting native rangeland conversion to fire-prone, annual grass-dominated communities and maintaining the ecosystem services provided by perennial plants. The herbicide indaziflam (Rejuvra, Bayer, Leverkusen, Germany) may allow managers to selectively deplete annual grass seed banks in plant communities that continue to support desirable perennial vegetation, but nontarget impacts are difficult to assess in the small plots typically used in herbicide trials, and the potential for impacts to the seed bank is not well understood. To assess the potential for nontarget impacts resulting from indaziflam treatment, we used modified-Whittaker multiscale vegetation plots to compare diversity (species accumulation) in three treatment plots (73 g ai ha–1 indaziflam) and three control plots in sagebrush-grasslands near Pinedale, Wyoming that are invaded by cheatgrass (Bromus tectorum L.). We also assessed the density and richness of shallow (0- to 1-cm depth) and deep (1- to 5-cm depth) germinable seed banks in treatment and control plots by tracking seedling emergence from seed bank samples during a 20-wk greenhouse study. Vegetation data and seed bank samples were collected during the third growing season after treatment. Treatment did not impact aboveground species diversity, but this contrasted with the results of the seed bank assay; shallow and deep native seed bank density and shallow native seed bank richness were significantly reduced in treatment plots. All impacted species were detected in the aboveground plant community in treatment plots after herbicide application, suggesting that reduced native annual abundance may be temporary. Considering the potential for nontarget impacts to seed banks will help land managers accurately assess trade-offs when making treatment decisions.