Because visibility bias can confound attempts to detect changes in abundance, we evaluated factors that affect visibility rate in aerial surveys of wintering waterfowl. We placed waterfowl decoys in 32 2- x 0.25-kin strip transects in the Mississippi Alluvial Valley (MAV) during February 1990 and 1991 and observed the decoys under different experimental conditions. Visibility rate was influenced (P 0.10) by habitat, transect width, and decoy group size. We simulated variation in use of habitat and found that changes in use between open and wooded wetlands would cause changes in visibility rate and affect the power to detect a change in abundance. The effect of changes in visibility rate on likelihood of detecting population change depended on the magnitude and direction of population change and precision of the population index. For transect surveys of wintering ducks in the MAV we recommend reducing transect width from 250 to 150 m on each side of the aircraft and restricting comparisons between years when 70% of the population is likely to be distributed in open wetlands. Improved techniques for estimating abundance of wintering waterfowl are also needed so use of questionable population indices can be avoided.