In contrast to theoretical predictions of even adult sex ratios, males are dominating in many bird populations. Such bias among adults may be critical to population growth and viability. Nevertheless, demographic mechanisms for biased adult sex ratios are still poorly understood. Here, we examined potential demographic mechanisms for the recent dramatic shift from a slight female bias among adult eider ducks (Somateria mollissima) to a male bias (about 65% males) in the Baltic Sea, where the species is currently declining. We analysed a nine-year dataset on offspring sex ratio at hatching based on molecularly sexed ducklings of individually known mothers. Moreover, using demographic data from long-term individual-based capture-recapture records, we investigated how sex-specific survival at different ages after fledgling can modify the adult sex ratio. More specifically, we constructed a stochastic two-sex matrix population model and simulated scenarios of different survival probabilities for males and females. We found that sex ratio at hatching was slightly female-biased (52.8%) and therefore unlikely to explain the observed male bias among adult birds. Our stochastic simulations with higher survival for males than for females revealed that despite a slight female bias at hatching, study populations shifted to a male-biased adult sex ratio (> 60% males) in a few decades. This shift that was driven by prime reproductive-age individuals (≥5-year-old), with sex-specific survival of younger age classes playing a minor role. Hence, different age classes contributed disproportionally to population dynamics. We argue that an alternative explanation for the observed male dominance among adults - sex-biased dispersal - can be considered redundant and is unlikely, given the ecology of the species. The present study highlights the importance of considering population structure and age-specific vital rates when assessing population dynamics and management targets.