Why do some restored ecosystems persist for centuries while others are quickly converted to alternative land uses or land covers? We propose that restored ecosystems have a temporal dimension that is variable, often finite, and likely predictable to some extent based on attributes of stakeholders, environment, and governance. The longevity of a restored ecosystem carries strong implications for its capacity to support biodiversity and provide ecosystem services, so an emerging challenge for restoration ecology is to predict the circumstances under which restored ecosystems persist for longer or shorter periods of time. We use a case study in tropical forest restoration to demonstrate one way that restored ecosystem longevity can be approached quantitatively, and we highlight opportunities for future research using restoration case study repositories, practitioner surveys, and historical aerial imagery. Much remains to be learned, but it is likely that decision-makers and practitioners have considerable leverage to increase the probability that restored ecosystems persist into the future, extending the benefits of contemporary restoration initiatives.