Simulated grazing techniques were used to investigate livestock impacts on structural and vegetation characteristics of streambanks in central Idaho, USA. The treatments, continued over two years, consisted of no grazing, simulated moderate early summer grazing, simulated moderate mid-summer grazing, and simulated heavy season-long grazing. The moderate treatments depressed the streambank surface about 3 cm, while the heavy season-long treatment resulted in an 11.5-cm depression. There were no differences between the no-grazing and moderate-grazing treatments for change in stream width, bank angle, bank retreat, or root biomass. The heavy season-long treatment, however, produced significant changes in these variables. The amount of foliage biomass (i.e., kg ha⁻¹) removed by treatment was similar between the two years of study for the moderate treatments. The foliage removed from the heavy season-long treatment plots greatly decreased in the second year as plant growth decreased. Ten months after the last treatment application, the average spring foliage growth was 20–43% lower on the moderate treatment plots and 51–87% lower on the heavy season-long treatment plots than on the untreated control plots.