The relationships between wetland water conditions and breeding numbers of Mallard (Anas platyrhynchos), Northern Pintail (A. acuta), Blue-winged Teal (A. discors), and Northern Shoveler (A. clypeata) during May of 1992-1995, were examined on twelve study areas in the eastern Prairie Pothole Region. Data were collected on water levels (by wetland class [temporary, seasonal, semipermanent]), pond density (density of wet basins), and numbers of indicated pairs for each species from weekly roadside transect surveys. Comparison of models relating duck numbers to wetlands using Akaike's Information Criterion indicated that measures of water condition generally were of similar value for explaining duck numbers. The model containing effects of semipermanent wetland water levels was among the best in explaining duck numbers for all species. Inclusion of temporary and seasonal wetland water levels in models for Mallard and Northern Pintail was not strongly supported by the data. Variation in duck numbers was much higher among areas than among years. Water conditions accounted for nearly all among-year variation for individual sites, but a large proportion of residual variation remained unexplained. Water condition measures (excluding spatial and temporal factors) explained 9-49% of variation in duck numbers, leaving 51-91% unexplained. Comparisons of these results to those of studies conducted at local or regional scale indicated that the relationship between duck numbers and pond numbers varied with scale, and suggested that other area-related factors should be considered at smaller landscape scales.