Monitoring demographic response over time is valuable for understanding population dynamics of endangered species. We quantified the variation in survival patterns for three small isolated island populations of endangered waterfowl in the Hawaiian Archipelago. Laysan Teal Anas laysanensis were individually marked and the fate of 1,150 individuals were followed from different cohorts among the two reintroduced (Kure and Midway Atolls) and the single relict (Laysan Island) populations for time series of 4, 10 and 15 years respectively. We applied a non-parametric Kaplan-Meier estimator to describe variation between the populations in survival for different cohorts. For Laysan Island and Midway Atoll, we used log-rank tests to determine the effects of cohort, island and sex on survival.  Birds in the Laysan Island population had significantly higher survival than those in the Midway population, and males had higher survival than females in both populations. The proportion of females surviving at Midway Atoll was 40% lower than for females on Laysan Island at year 5. The oldest bird observed from Laysan Island was at least 15.5 years old and had been ringed as an adult. The Kure Atoll founder cohort (n = 28) had 100% survival 18 months post-release, but this dropped by 39% during the first avian botulism type C outbreak. Ten of twenty-eight founders and a population of 60–70 birds persisted on Kure Atoll in 2020. We summarised mortality records to generate hypotheses to explain the cause-specific mechanisms driving the observed survival differences. Mortality data showed that the survival differences between islands in Laysan Teal survival was driven by chronic epizootics of avian botulism type C at Midway and Kure Atoll.