Information regarding microhabitat, here defined as small-scale vegetation structure, is often useful in predicting use of habitat by birds. Quantifying microhabitat, however, is expensive and labor intensive compared to assessment of habitat at a larger scale, possibly from remotely sensed imagery. To assess the importance of microhabitat information in constructing predictive models of habitat occupancy, we compared occupancy models built on the basis of macro- and microhabitat together and separately. We based our models on counts of wintering migratory bird species and vegetation surveys within Tuskegee National Forest, Alabama, completed during winter 2009. Models built from macrohabitat data only outperformed models built from microhabitat data only for five of the six species analyzed. However, the best model for every focal species included both macro- and microhabitat covariates. Pine forests—excluding plantation—were the only land-cover classification important to our focal species, and measures of density of vegetation were important in predicting occupancy. Our results suggest that migrants wintering at our study site select habitat at multiple scales—specializing in certain types of cover and then preferring specific structural aspects of vegetation within them. We conclude that microhabitat information is important for inference into use of habitat by wintering migratory birds.