In ecosystems with alternative stable states, restoration success can be thought of as overcoming the resilience of an undesirable state to promote an alternative state that yields greater ecosystem services. Since greater resilience of undesirable states translates into reduced restoration potential, quantifying differences in resilience can enhance restoration planning. In the context of shrub-encroached rangeland restoration, shrubland resilience is the capacity of a woody vegetated state to absorb management interventions designed to produce a more desirable grass-dominated state, and remain within its current regime. Therefore, differences in the resilience of a state can be quantified in a relative sense by measuring whether a state switches to an alternate state following perturbation or remains in its current stability domain. Here we designed an experimental manipulation to assess the contribution of soils to differences in the relative resilience of a shrub-invaded state. In this large-scale experiment, we repeated perturbations across a gradient of soil textures to inform restoration practitioners of differences in the relative resilience of shrubland occurring on different soil types to common rangeland restoration practices. On each soil type, we compared the relative ability of the shrubland state to withstand chemical and mechanical brush control treatments, commonly employed in this study region, to untreated controls. While the shrubland community composition did not differ prior to the study, its capacity to absorb and recover from brush removal treatments depended on soil type. Shrubland resilience to chemical and mechanical brush removal was highest on coarse soils. On these soils, brush removal temporarily restored grassland dominance, but woody plants quickly regained pretreatment levels of dominance. However, shrublands on fine soils did not recover following treatments, continuing to be grass-dominated for the duration of the study. This study highlights a simple approach for prioritizing restoration actions by mapping the locations of different soil attributes that support shrub-dominated states with differing levels of resilience to brush control. This experimental approach provides a basis for operationalizing resilience in restoration and prioritizing management actions across a range of environmental conditions, which is critical given the economic constraints associated with broad-scale mechanical and chemical interventions for rangeland restoration.