To produce better habitat models, we analyzed a 1977 breeding bird data set with recently developed techniques. Our generalized linear models for specific bird species produced precise descriptions of bird–vegetation relationships, using visually obvious habitat features (e.g., grass, broad-leaved shrubs, and conifers of differing heights). This was, in general, confirmed by canonical correspondence analysis, which related the entire bird community to the vegetation across the successional stages. Nevertheless, our analyses produced some results that contradicted our expectations; ground-nesting species negatively associated with bunch grasses (species that grow in clumps within the Agropyron, Festuca, and Stipa genera), and cavity-nesting species negatively associated with stumps and snags. We speculate that those contrary relationships resulted from nest sites not being a limiting resource. We conclude that visually obvious vegetation features are sufficient to describe most breeding bird–habitat relationships. Further, appropriate landscape management for bird biodiversity requires that all successional stages are present.