The Aleutian cackling goose (Branta hutchinsii leucopareia) was listed as endangered in 1967, downgraded to threatened in 1990, and removed from protection under the endangered species act in 2001. Accurate determination of population status continues to be a priority for management agencies because of the population's past listing as endangered, the species status as a game bird, and because population expansion has resulted in increased crop damage complaints. We compared estimates of marked Aleutian goose abundance from available mark-resight data during 1996–2012 using capture–recapture Jolly–Seber (POPAN formulation) and robust design (closed capture) models. We derived an estimate of Aleutian goose abundance by expanding an estimate of marked-bird abundance by the ratio of total to marked geese. Robust design models provided strong evidence of even-flow and Markovian temporary emigration from observable to unobservable states between annual sampling periods, and POPAN estimates were biased as a result of this movement. The observable population increased in abundance from 790 in 1975 to 134,703 (SE = 10,866, 95% CI = 113,405–156,002) geese in 2012. However, total (observable and unobservable) population size was 216,564 geese based on an estimated probability (0.622) of being available for observation. Annual survival probability was 0.80 (SE = 0.05). Current data collection allows estimation of annual observable abundance with a coefficient of variation of ≤8%. Robust design offers a means of estimating abundance of partially marked wildlife populations in the presence or absence of temporary emigration, which can result from partial study area access and is likely a common situation for many populations. Mark-resight is a cost-effective approach (compared to capture–recapture studies and aerial surveys) that can be applied to successfully estimate demographic parameters of other populations. © 2013 The Wildlife Society.