We compared soil-surface penetration resistance and abundance of ant mounds at 12 western Great Basin sites (composed of 19 plots) either grazed by feral horses (Equus caballus) or having had horses removed for the last 10–14 years. Across this broad spatial domain (3.03 million ha), we minimized confounding due to abiotic factors by selecting horse-occupied and horse-removed sites with similar aspect, slope, fire history, grazing pressure by cattle (minimal to none), and dominant vegetation (Artemisia tridentata). During both 1997 and 1998, we found 2.2–8.4 times greater abundance of ant mounds and 3.0–15.4 times lower penetration resistance in soil surfaces at horse-removed sites. In 1998, thatched Formica ant mounds, which existed predominately at high elevations, were 3.3 times more abundant at horse-removed sites, although abundance varied widely among sites within treatments. Several types of analyses suggested that horses rather than environmental variability were the primary source of treatment differences we observed in ecosystem components. Tests of several predictions suggest that alterations occurred through not only direct effects, but also indirect effects and potentially feedback loops. Free-roaming horses as well as domestic grazers should be considered in conservation planning and land management in the Great Basin, an ecoregion that represents both an outstanding conservation opportunity and challenge.