Over the last century, many grasslands in the Chihuahuan Desert have converted to shrublands dominated by creosotebush (Larrea tridentata). The loss and spatial isolation of perennial grasslands has led to declines of grassland-dependent wildlife species. Grassland restoration efforts have been dominated by the application of herbicide to control creosotebush over large spatial scales. However, we have a limited understanding of how landscape-scale restoration has affected wildlife species. We examined whether restoration treatments in southern New Mexico have increased the abundance of the banner-tailed kangaroo rat (Dipodomys spectabilis), a keystone rodent that forages on perennial grasses. We selected 21 sites that were treated with herbicide from 7-29 years ago, and we compared the abundance of D. spectabilis on treated areas with paired reference sites that were dominated by creosotebush. To examine potential constraints on restoration success, we also tested whether the response of D. spectabilis depended on time since treatment, treatment area, and broader landscape connectivity. The abundance of D. spectabilis was greater on sites treated with herbicide than on paired references. However, the response of D. spectabilis depended on time since treatment and spatial connectivity. Abundance was related positively to treatment age, suggesting a lagged response of D. spectabilis to restoration. Abundance also increased with connectivity to other restoration sites, indicating that restoration success can depend on landscape context. Abundance was particularly high at sites that were near old treatments. The area of treatments did not have a strong effect on abundance. Dipodomys spectabilis greatly affects ecosystem functioning, and it improves habitat for other wildlife species by constructing large mounds that provide cover. Overall, our results suggest that recolonization by D. spectabilis at restoration sites may depend on the age and spatial configuration of treated areas.