We used band recovery data from grey ducks (Anas superciliosa) banded in New Zealand between 1957 and 1974 to test 2 null hypotheses: (1) hunting mortality is completely additive to natural sources of mortality, and (2) hunting mortality is completely compensated by changes in natural mortality. We modeled annual survival as a function of survival in the absence of hunting and the probability of death from hunting. The complete compensation hypothesis was rejected, but we were unable to reject the completely additive hypothesis. There was no evidence of sex- or age-specificity for the relationship between kill rate and annual survival rate. We used simulated data to evaluate model performance. Parameter estimates were unbiased despite the inclusion of estimates that lay outside the bounds of the parameter space, although model-based variance estimates were consistently less than empirical variances. Our results imply that harvest-restrictions may be useful in effecting change in annual survival rates of grey ducks.