Service use annual estimates of the number of ponded wetlands to estimate duck production and establish duck hunting regulations. Sampling techniques that minimize bias may provide more reliable estimates of annual duck production. Using a wetland geographic information system (GIS), we estimated number of wetlands using standard counting protocol with belt transects and samples of square plots. Estimates were compared to the known number of wetlands in the GIS to determine bias. Bias in transect-derived estimates ranged from +67-87% of the known number of wetlands, compared to bias of +3-6% in estimates from samples of 10.24-km2 plots. We recommend using samples of 10.24-km2 plots stratified by wetland density to decrease bias.