Although the impacts of livestock and human activities on riparian zones and associated wildlife have been well documented, little is known about the impacts that browsing by large native ungulates such as elk and moose may have. In the northern Yellowstone area, some willow stands experience intense browsing by elk and moose whereas others experience medium or very low amounts of browsing. The objectives of this study were (1) to compare the species and densities of birds among willow stands that have experienced different intensities of browsing by native ungulates, (2) to measure the relationship between five species of birds and aspects of habitat structure, and (3) to develop and evaluate predictive models that relate presence or absence of the five species to habitat characteristics. In 1989 and 1990, I measured densities of nesting songbirds and aspects of habitat structure in eight large willow stands that have experienced different intensities of browsing. The densities of five focal species (Common Yellowthroat, Lincoln's Sparrow, Warbling Vireo, Wilson's Warbler, and Yellow Warbler) varied considerably among sites. Only two sites had all five species and only one species--the Lincoln's Sparrow'was found in all eight sites. The proportion of severely browsed willows in the eight sites ranged from 3.5% to 100%. The nonlinear relationship between total bird densities and frequency of severe browsing suggests that birds have a threshhold of tolerance for browsing, beyond which bird numbers and total numbers of species drop. Principal Components Analysis of 14 habitat variables indicates that the study sites varied in terms of distances between shrubs, shrub heights, height heterogeneity, foliage density at various height intervals, and frequency of severely browsed willows. Browsing does appear to affect the assemblages of breeding birds in these sites, but site- and landscape-level factors such as food abundance, willow species composition, hydrology, type and gradient of adjacent community, and riparian zone width and elevation also play important roles. such variables should be incorporated into future predictive models to improve model performance. (82 pages)