Wetland basins in the Prairie Pothole Region of the U.S. are commonly modified by excavation (e.g., roadside ditches, stock dugouts), partial drainage (ditching), and diking. Differences in the distribution of modified wetlands may affect the predictive accuracy of waterfowl survey data if such wetlands are not distributed randomly in the landscape and if waterfowl are not distributed equally among them. We used data collected on thirty-eight 40-km2 plots in North Dakota to examine the distribution of modified basins relative to roadside transects and their use by five species of dabbling ducks in 1995. The 800-m-wide transects were subdivided into an inner 400-m transect, centered on the road, and the remaining outer transect area. We compared the distribution of modified and natural wetland basins among three sample areas: 1) the inner 400-m wide roadside transect area, 2) the outer transect area, and 3) the remaining area within the 40-km2 plot that was outside of the transects (outer plot). Duck use was compared between the two transect areas. The plots contained 20,582 basins, of which 88.5% were unmodified, 7.5% were excavated, 3.7% were partially drained, and 0.2% were diked. Nearly all excavated temporary (89%) and seasonal (90%) basins occurred in the inner transect area, reflecting the high proportion of basins that would be defined as roadside ditches. Excavated semipermanent basins were more evenly distributed among the outer plot and two transect widths; these basins often were dugouts but also included roadside ditches. Partially drained and diked basins also were fairly evenly distributed among the three sample areas. Semipermanent basins had greater use by mallards (Anas platyrhynchos) and northern pintails (A. acuta) when they were partially drained than when they were excavated or unmodified; pintails also had greater use of partially drained seasonal basins. Use of wetland basins by gadwall (A. strepera), blue-winged teal (A. discors), and northern shovelers (A. clypeata) did not differ among water regimes or modification. We found no evidence to indicate that duck numbers determined from standard 400-m-wide roadside transects were biased relative to the larger landscape. However, pond counts derived from such transects were biased. Correlations of duck numbers to pond counts that exclude ditches or temporary basins would poorly reflect the response of ducks to available water.