The use of tissue stable isotope values to track animal movements is well-established but independent tests of existing models and assignment accuracy at coarse spatial scales and for non-model species are rare. Assignment error associated with an existing model for distinguishing origins of North American Mallards (Anas platyrhynchos) on the basis of feather δ³⁴S, δD, δ¹⁵N and δ¹³C values was evaluated. Prior information about regional movements reported in Mallard mark-recapture studies was accounted for, and new models generated and applied to independent data obtained from flightless, juvenile Mallard and Lesser Scaup (Aythya affinis). Assignment error of an existing model was reduced from 14 to 8% by incorporating prior information about mallard movement rates between Pacific and Mid-continent Flyways (range 0 to < 1%) in Flyway-specific models. The revised Mid-continent model correctly assigned 86% of 70 flightless hatch-year Mallards captured at breeding areas in central North America during 2005 to origin. Assignment accuracy ranged from 76 to 91% over five 1° latitude increments where Boreal Forest and Prairie regions were increasingly separated from a focal Aspen Parkland region; here, separating Boreal Forest and Prairie regions from the Aspen Parkland by 4° latitude provided the highest relative assignment success rates among all three regions. Estimated origins of known-source scaup were accurate (14 of 16) for birds with feathers grown in Aspen Parkland but not (0 of 5) for those grown in the Canadian North-western Boreal Forest. The results support the use of multiple feather-isotopes for assigning natal origins of birds at broad geographic scales but evaluation of factors affecting species-specific isotope differences is required for application of existing models to other species.