The data set consists of data collected in 1995, 1996, and 1997 in wetlands in the Prairie Pothole Region of North Dakota and South Dakota. The data were summarized and used in the analysis for two publications in peer-reviewed journals. The data consists of one data set that includes bird abundance and richness data, wetland class (temporary, seasonal, semipermanent, permanent, alkali), and wetland type (private or federal). In conjunction with surveys of wetland birds, the following wetland variables also were recorded: wetland size, date, local-level habitat variables (percent coverage of open water, emergent vegetation, wet meadow, and shoreline/mudflat), and landscape-level habitat variables (percent cover of grassland, hayland, planted cover, cropland, woodland, rights-of-way, barren land, wetland, shrubland, feedlot, and odd areas). The first publication is entitled: The relative importance of wetland area versus habitat heterogeneity for promoting species richness and abundance of wetland birds. In the first publication, we examined the relationship between habitat heterogeneity and species richness after accounting for habitat area in glacially formed wetlands in the Prairie Pothole Region in North America. We also assessed whether this relationship holds true 1) for richness of groups of similar species, 2) for species abundances, and 3) at both local within-wetland and landscape scales of habitat heterogeneity. The second publication is entitled: Regional modeling of habitat associations for wetland-obligate birds in the Prairie Pothole Region and Great Lakes basin. In the second publication, we developed Poisson models of species density for four species of secretive marsh birds (Pied-billed Grebe [Podilymbus podiceps], Virginia Rail [Rallus limicola], Sora [Porzana carolina], and American Bittern [Botaurus lentiginosus]) for wetlands in the Prairie Pothole Region (this data set) and wetlands in two other regions (inland Great Lakes and coastal Great Lakes). We used adjusted pseudo-R2 values to compare the amount of variation explained by each model when it was applied to data collected in its region and to data collected in the other two regions.