Seed dispersal dynamics strongly affect plant community assembly in restored annual grass—infested ecosystems. Modifying perennial grass seeding rates and frequency may increase perennial grass establishment, yet these impacts have not yet been quantified. To assess these effects, we established a field experiment consisting of 288 plots (1 m²) in an eastern Oregon annual grass—dominated shrubsteppe ecosystem. In this study, the amount, timing, and frequency of perennial grass seeding events, soil moisture availability, and annual grass seed bank density were manipulated. We found that more frequent perennial grass seeding events combined with high perennial grass seeding rates produced the highest perennial grass density and biomass 2 years following seeding. However, we also found that if annual grass seed density was 1500 seeds · m^-2 or higher, perennial grass density and biomass decreased, regardless of seeding strategy. Because of this finding, it appears that a threshold is crossed between 150 and 1500 annual grass seeds · m^-2. Adding water in the first growing season initially facilitated perennial grass establishment but only produced higher perennial grass density following the second growing season when annual grass density was lowest. Assessing the existing annual grass seed bank prior to seeding can likely forecast restoration outcomes because high annual grass seed densities likely interfere with and reduce perennial grass recruitment. In addition, if annual grass seeding density is 1500 seeds · m^-2 or lower, modifying the temporal patterns of perennial grass seed arrival will increase the likelihood that a perennial grass seed finds a safe-site.