AbstractAdaptive management of harvested waterfowl requires accurate estimations of demographic parameters. These must also be representative of the targeted population. In the greater snow goose, all demographic parameters so far have been estimated from long‐term banding conducted at a single nesting colony in the Arctic, Bylot Island, where 15% of the population breeds. We used data from a second banding program conducted on Ellesmere Island, 800 km north of Bylot Island and near the northern limit of the breeding range, to compare adult survival between these 2 breeding sites over the period 2007–2011. This allowed us to determine the representativeness of demographic parameters estimated from the Bylot colony. We used a multi‐event capture‐recapture model combining recaptures, resightings of neckbanded birds, and recoveries on a seasonal basis, which allowed us to test specifically for differences in survival during the migration periods. Despite differences in migration distance (20% longer for Ellesmere Island) and environmental conditions, survival rate of birds from these 2 breeding sites were similar in all seasons. Annual survival ranged from 0.72 to 0.79. This apparent absence of a cost of migration on survival may be explained by the canalization hypothesis: variance in adult survival of the greater snow goose, a long‐lived species, caused by environmental factors may have been reduced because of selection pressure on this trait, which is closely linked to fitness. The absence of spatial variation in adult survival suggests that the extrapolation of survival parameters estimated from the Bylot Island colony to the entire population may be valid. © 2015 The Wildlife Society.