Invasive weed species in California's rangelands can reduce herbaceous diversity, forage quality and wildlife habitat. Small-scale studies (5 acres or fewer) have shown reductions of medusahead and yellow starthistle using prescribed grazing on rangelands, but little is published on the effects of pasture-scale (greater than 80 acres) prescribed grazing on weed control and plant community responses. We report the results of a 6-year collaborative study of manager-applied prescribed grazing implemented on rangeland that had not been grazed for 4 years. Grazing reduced medusahead but did not alter yellow starthistle cover. Medusahead reductions were only seen in years that did not have significant late spring rainfall, suggesting that it is able to recover from heavy grazing if soil moisture is present. Later season grazing appears to have the potential to suppress medusahead in all years. In practice, however, such grazing is constrained by livestock drinking water availability and forage quality, which were limited even in years with late spring rainfall. Thus, we expect that grazing treatments under real-world constraints would reduce medusahead only in years with little late spring rainfall. After 10 years of grazing exclusion, the ungrazed plant communities began to shift, replacing medusahead with species that have little value, such as ripgut and red brome.