We experimentally studied the effects of avian predation on grasshopper abundance in western North Dakota during the summers of 1988 and 1989. Grasshopper densities in 15 ° 15 m plots from which birds were excluded (NO BIRDS) were compared with similar—sized plots where birds were allowed to forage (BIRDS). Plots were established in early June at 16 sites (8 per year), and grasshopper densities were estimated from hoop counts in NO BIRDS and BIRDS plots at 2—wk intervals until the end of July. There were significantly more grasshoppers in NO BIRDS plots than in BIRDS plots in 1989 (P < .0001), but not in 1988 (P = .137). The difference between treatments in 1989 was first detected 6 wk after the exclosures were erected. Between years there were no differences in initial grasshopper abundance in all treatments (P > .388). After the final hoop count each year, sweep—net sampling was also used to estimate grasshopper densities. By this method, average grasshopper density in late July was 26% and 37% lower in BIRDS plots than in NO BIRDS plots in 1988 and 1989, respectively. Average length, total biomass of grasshoppers, species richness, and species diversity, however, did not differ between the treatments.In 1988, 2 of 15 grasshopper species were significantly more abundant in the NO BIRDS plots. There was no difference between the treatments among 16 species identified in the 1989 samples, but the power of our tests to detect differences for individual species was low. Our results support the hypothesis that avian predation reduces insect populations at low and moderate densities.