A recently developed radar-based technique permitted empirical re-evaluation of the established but poorly supported theory that migrating North American waterfowl (Anatidae) use landscape features such as rivers as leading lines. Ducks departing the Illinois River Valley in the autumn of each of 15 years travelled SSE with a mean track that was 68° different from the 220° course of the Illinois River (P = 0.001). We conclude that leading lines were unimportant navigation aids for ducks leaving this major stopover site in autumn and suggest that rivers have less effect on the spatial course of duck migration than previously thought. Timing of departures was examined in a representative subset of 8 years and found to be consistent, with a mean start time of 44 min after civil sunset. (c) 2014 British Ornithologists' Union.