Strip transects located along roads are commonly used to estimate waterfowl populations and characterize associated wetland habitat. We used data collected in May and early June, 1995, on forty-five 40-km² plots in North Dakota to evaluate bias of 800-m and 400-m wide roadside transects for sampling wetlands relative to a larger (40-km²) scale and to compare duck abundance at the two widths. Densities of all basins combined and of seasonal basins considered alone were biased high for both transect widths, but mean bias did not differ from zero for temporary or semipermanent basins. Biases did not occur when excavated seasonal and temporary basins (i.e., road ditches) were excluded from the sample. Mean basin density was higher for the inner (400-m) transect width than for the outer transect width (area remaining of the 800-m transect, outside of center 400-m width) for all basins combined and for seasonal and temporary basins. We detected an area-related bias in the occurrence of basins in transects: smaller basins (0.08–1.6 ha) were over-represented in transect samples by 2.9–6.5%, and larger basins (≥11 ha) were under-represented in 800-m transects by 7.3% and in 400-m transects by 16.3%. We compared the distribution of ducks relative to water conditions in the inner and outer transect widths to evaluate whether they were affected by proximity to the road. Mallards (Anas platyrhynchos L.), northern pintails (A. acuta L.), and gadwall (A. strepera L.) responded to water conditions equally in the inner and outer transect widths, but northern shovelers (A. clypeata L.) and blue-winged teal (A. discors L.) responded more strongly to wetlands on the inner than the outer transect width, indicating that estimates of these species would be higher from a 400-m wide transect than from an 800-m wide transect. Differences in an adjustment index, used to account for the portion of basin obscured from view, were highly variable between inner and outer transect widths but did not indicate that use of wider transects was hampered by visibility. Biases of transect sampling need to be carefully considered when extrapolating wetland basin or duck densities from transects to larger landscape scales.