The study evaluates the potential for bird species assemblages to serve as indicators of biological integrity of rangelands in the Great Basin in much the same way that fish and invertebrate assemblages have been used as indicators in aquatic environments. Our approach was to identify metrics of the bird community using relatively simple sampling methods that reflect the degree of rangeland degradation and are consistent over a variety of vegetation types and geographic areas. We conducted the study in three range types (i.e., potential natural plant community types) in each of two widely separated areas of the Great Basin: south-eastern Idaho (sagebrush steppe range types) and west-central Utah (salt-desert shrub range types). Sites were selected in each range type to represent three levels of grazing impact, and in Idaho included sites modified for crested wheatgrass production. Birds were sampled by point counts on 9 100-m radius plots at 250-m spacing on each of 20 sites in each area during the breeding season. In sagebrush-steppe, 964 individuals in 8 species of passerine birds were used in analyses. Five metrics were significantly related to impact class, both when analyzed within range type and when analyzed with all range types combined. Species richness, relative abundance of shrub obligate species, and relative abundance of Brewer's sparrow were generally lower for the higher impact classes, whereas the reverse was true for dominance by a single species and for relative abundance of horned larks. In contrast, total number of individuals did not differ significantly as a function of impact class. In salt-desert shrub, a total of 843 birds in 4 species were included in analyses, 98% of which were horned larks. None of the metrics identified above was significantly related to impact class. Two metrics for breeding birds in sagebrush steppe (species richness and dominance) showed little overlap between values for the extremes of impact class, and thus they have potential as indicators of biological integrity. However, the sensitivity of these metrics appears to be greatest at the high impact end of the spectrum, which suggests they may have limited utility in distinguishing between sites having light and moderate impact.