Cheatgrass (Bromus tectorum L.) is an invasive annual grass that creates near-homogenous stands in areas throughout the Intermountain sagebrush steppe and challenges successful native plant restoration in these areas. A clipping experiment carried out at two cheatgrass-dominated sites in eastern Oregon (Lincoln Bench and Succor Creek) evaluated defoliation as a potential control method for cheatgrass and a seeding preparation method for native plant reseeding projects. Treatments involved clipping plants at two heights (tall  =  7.6 cm, and short  =  2.5 cm), two phenological stages (boot and purple), and two frequencies (once and twice), although purple-stage treatments were clipped only once. Treatments at each site were replicated in a randomized complete block design that included a control with no defoliation. End-of-season seed density (seeds · m⁻²) was estimated by sampling viable seeds from plants, litter, and soil of each treatment. Unclipped control plants produced an average of approximately 13 000 and 20 000 seeds · m⁻² at Lincoln Bench and Succor Creek, respectively. Plants clipped short at the boot stage and again 2 wk later had among the lowest mean seed densities at both sites, and were considered the most successful treatments (Lincoln Bench: F_6,45  =  47.07, P < 0.0001; Succor Creek: F_6,40  =  19.60, P < 0.0001). The 95% confidence intervals for seed densities were 123–324 seeds · m⁻² from the Lincoln Bench treatment, and 769–2 256 seeds · m⁻² from the Succor Creek treatment. Literature suggests a maximum acceptable cheatgrass seed density of approximately 330 seeds · m⁻² for successful native plant restoration through reseeding. Thus, although this study helped pinpoint optimal defoliation parameters for cheatgrass control, it also called into question the potential for livestock grazing to be an effective seed-bed preparation technique in native plant reseeding projects in cheatgrass-dominated areas.