The U.S. Fish and Wildlife Service in an effort to improve and standardize methods for estimating waterfowl production tested a new technique in the four-county Arrowwood Wetland Management District (WMD) for three years (1982-1984). On 14 randomly selected 10.36 km2 plots, upland and wetland habitat was mapped, classified, and digitized. Waterfowl breeding pairs were counted twice each year and the proportion of wetland basins containing water was determined. Pair numbers and habitat conditions were entered into a computer model developed by Northern Prairie Wildlife Research Center. That model estimates production on small federally owned wildlife tracts, federal wetland easements, and private land. Results indicate that production estimates were most accurate for mallards (Anas platyrhynchos), the species for which the computer model and data base were originally designed. Predictions for the pintail (Anas acuta), gadwall (A. strepa), blue-winged teal (A. discors), and northern shoveler (A. clypeata) were believed to be less accurate. Modeling breeding period dynamics of a waterfowl species and making credible production estimates for a geographic area are possible if the data used in the model are adequate. The process of modeling the breeding period of a species aids in locating areas of insufficient biological knowledge. This process will help direct future research efforts and permit more efficient gathering of field data.