Population forecasts are used to manage threatened and endangered wildlife populations. Natural resource managers use these predictions to anticipate future changes in species abundance, assess extinction risk, and prioritize management interventions. Inaccurate forecasts may lead to erroneous interventions or inefficient uses of limited resources including funding and personnel. Despite the widespread adoption of population forecasts over the last four decades, there have been few efforts to assess the historical performance of these predictions. This project will use the growing number of long-term monitoring datasets collected by research scientists and state and federal agencies to assess the forecast performance of population models. Findings will inform management strategies for threatened populations by identifying the types of data and models that generate accurate forecasts. Outcomes of this project include improved guidance for natural resource managers on effective monitoring strategies for threatened populations, and the development of a framework that can be applied to evaluate other historical ecological forecasts. The research will train the next generation of scientists with modeling and programing skills, handling and development of databases, and the development of AI-ready databases for the scientific community. Population ecologists have been making predictions on the risk of population decline and extinction for almost 40 years. While there has been some past work evaluating forecast ability in stable populations, most assessments of population viability forecasts have been through indirect methods, thus, there is little empirical evidence assessing the long-term accuracy of these forecasts. This project will apply a retrospective approach to assess the reliability of population predictions by developing a publicly available database of historical population viability forecasts linked to updated monitoring data of vertebrates, invertebrates,