Invasive non-native species pose a large threat to restoration efforts following large-scale disturbances. Bromus tectorum (cheatgrass) is a non-native annual grass in the western U.S. that both spreads quickly following fire and accelerates the fire cycle. Herbicide and seeding applications are common restoration practices to break the positive fire-invasion feedback loop and recover native perennial species, but their interactive effects have infrequently been tested at the landscape-scale and repeated in time to encourage long-lasting effects. We determined the efficacy of repeated post-fire application of the herbicide imazapic and seeding treatments to suppressBromus abundance and promote perennial vegetation recovery. We found that the selective herbicide reduced Bromus cover by ~30 % and density by >50 % across our study sites, but had a strong initial negative effect on seeded species. The most effective treatment to promote perennial seeded species cover was seeding them alone followed by herbicide application 3 years later when the seeded species had established. The efficacy of the treatments was strongly influenced by water availability, as precipitation positively affected the density and cover of Bromus; soil texture and aspect secondarily influenced Bromus abundance and seeded species cover by modifying water retention in this semi-arid region. Warmer temperatures positively affected the non-native annual grass in the cool-season, but negatively affected seeded perennial species in the warm-season, suggesting an important role of seasonality in a region projected to experience large increases in warming in the future. Our results highlight the importance of environmental interactions and repeated treatments in influencing restoration outcomes at the landscape-scale.