Where stable source populations of at-risk species exist, translocation may be a reasonable strategy for re-establishing extirpated populations. However, the success rates of such efforts are mixed, necessitating thorough preliminary investigation. Stochastic population modeling can be a useful method of assessing the potential success of translocations. Here, we report on the results of modeling translocation success for the Hawaiian Common Gallinule (‘alae ‘ula; Gallinula galeata sandvicensis), an endangered waterbird endemic to the Hawaiian Islands. Using updated vital rates, we constructed a model simulating three existing extant (wild) source populations and a hypothetical recipient site on another island. We then projected the effects of six different translocation scenarios and sensitivity of the results to variation of three important demographic parameters on the probability of extinction (PE) of the reintroduced and donor populations. Larger translocations, of at least 30 birds, had low probability of extinction in the reintroduced population, but raised extinction risk of the smallest source population. Spacing out translocations in time (e.g., 10 birds translocated in total in three installments over nine years), led to lower PE than translocating all individuals at once (i.e., bulk translocations) for both the source and reintroduced populations. Brood size and hatch-year juvenile survival had a disproportionate impact on reintroduced population viability. Importantly, the reported juvenile survival rate is very near the threshold for population failure. This suggests that post-introduction and subsequent management of wetlands, particularly predator control, could be critical to reintroduction success. We recommend that individuals should be translocated from multiple, genetically distinct subpopulations to reduce the possibility of inbreeding depression. Based on this analysis, the recipient wetland should be sufficiently large that it can support at least 25 pairs of gallinules. Based on recent estimates of population densities on O‘ahu, such a wetland would need to be between 3.75-74.6 ha. Where stable source populations of at-risk species exist, translocation may be a reasonable strategy for re-establishing extirpated populations. However, the success rates of such efforts are mixed, necessitating thorough preliminary investigation. Stochastic population modeling can be a useful method of assessing the potential success of translocations. Here, we report on the results of modeling translocation success for the Hawaiian Common Gallinule (‘alae ‘ula; Gallinula galeata sandvicensis), an endangered waterbird endemic to the Hawaiian Islands. Using updated vital rates, we constructed a model simulating three existing extant (wild) source populations and a hypothetical recipient site on another island. We then projected the effects of six different translocation scenarios and sensitivity of the results to variation of three important demographic parameters on the probability of extinction (PE) of the reintroduced and donor populations. Larger translocations, of at least 30 birds, had low probability of extinction in the reintroduced population, but raised extinction risk of the smallest source population. Spacing out translocations in time (e.g., 10 birds translocated in total in three installments over nine years), led to lower PE than translocating all individuals at once (i.e., bulk translocations) for both the source and reintroduced populations. Brood size and hatch-year juvenile survival had a disproportionate impact on reintroduced population viability. Importantly, the reported juvenile survival rate is very near the threshold for population failure. This suggests that post-introduction and subsequent management of wetlands, particularly predator control, could be critical to reintroduction success. We recommend that individuals should be translocated from multiple, genetically distinct subpopulations to reduce the possibility of inbreeding depression. Based on this analysis, the recipient wetland should be sufficiently large that it can support at least 25 pairs of gallinules. Based on recent estimates of population densities on O‘ahu, such a wetland would need to be between 3.75-74.6 ha. Methods Reproductive rate data (HAGAVitalRates_9-10-23_Export) We acquired nest data from recent monitoring projects run through the state of Hawaii Department of Land and Natural Resources, Division of Forestry and Wildlife (DOFAW) on O‘ahu, and graduate dissertation work conducted at Hanalei National Wildlife Refuge on Kaua‘i (by BW). Nests on O‘ahu were located during routine weekly or biweekly surveys using an area-search survey. A team of 3–7 observers walked meandering transects with the goal of locating all nests in a given area. All nests were visually checked 2 times per week until hatching or failure. DOFAW nest monitoring continued throughout the annual cycle. A subset of Hawaiian Gallinule nests on O‘ahu was monitored from January through December 2020–2023. All nests were visually checked at least twice weekly, and a subset was monitored from January through December 2020–2023 using SPYPOINT Solar Dark (GG Telecom, Quebec, Canada) passive infrared cameras (trigger speed: 0.07 s) placed about 1 m from the nest, mounted on a 7.6 cm wide metal post 1.8 m long, fixed with a fully adjustable camera mount that allows a camera angle of 0–90. Cameras were programmed to take 2 images back-to-back immediately upon infrared motion activation. Cameras were programmed to take photos instantly for each activation (Instant setting recovery speed: 0.3 s). Cameras were checked weekly for battery life and SD card data retrieval and were removed either immediately after a nest wasconfirmed failed or after a nest was confirmed successful. A nest was considered successful if at least 1 egg hatched and was considered failed if the eggs all disappeared before the expected hatch date or if signs of predation (e.g., predator scat/tracks in the nest or destroyed eggs adjacent to the nest), flooding (e.g., intact eggs outside nest following an increase in water level or nest submerged under water), or abandonment (e.g., eggs cold to the touch in the morning, hot to the touch in the afternoon) were apparent. On Kaua‘i, nests were found by conducting systematic searches. In wetland units managed strictly for waterbirds, transects spaced 10 m apart were walked, while in taro that was grown on the refuge searches were done by walking the pond perimeter. Although Hawaiian Gallinules can nest year-round (Shallenberger 1977, Byrd and Zeillemaker 1981), searches were concentrated during the main breeding season. Nests also were found incidentally during regular activities by refuge staff and taro farmers. Nests on Kaua‘i were monitored with and without cameras (see Webber 2022 for details of monitoring and assessment of nest fates). All nests were checked every 3–5 d to monitor nest status; if the brood continued to use the nest after hatching and the camera was available, monitoring continued for brood survival data. Brood Survival Data (BroodDatabase_8-24-22) Due to some methodological differences in brood monitoring among datasets compiled in this study, we resampled data to a matching, lowest common temporal resolution. Brood data from Keawawa wetland (O‘ahu) were recorded via multiple daily surveys for the first 60þ d posthatch by a group of trained citizen science volunteers. Brood encounter data on Kaua‘i were collected based on 4 d encounter intervals, recording presence and number of chicks if the brood was detected on any day within the interval. All brood records in our Kaua‘i dataset were collected by BW, and they were monitored by surveying telemetered adults (see Webber 2022 for details). Territories with known nests were monitored starting at what was estimated to be mid-incubation, and visited at least once every 4 d. At James Campbell National Wildlife Refuge (O‘ahu), brood encounters were opportunistic. Except for data from Keawawa, most brood monitoring ended after the first month post-hatch. Based on these data formats, we reduced our combined data to 4 d intervals and the first 30 d post-hatch to avoid estimating parameters with a sparse dataset.