Reintroduction programs often face the challenge of sustaining multiple populations, each with unique demographic rates. Decision makers must determine how to allocate a finite number of captive-reared animals among these populations to achieve the fundamental objective of the program. We explored the optimal allocation of captive-reared Puerto Rican parrots (Amazona vittata) to 3 distinct populations: a struggling relict (population 1), a successfully reintroduced population (population 2), and a potential (new) population (population 3). We illustrate decision tradeoffs using 4 hypothetical scenarios of demographic performance created using the parrot literature and expert opinion and find the optimal decision by formulating the problem as a Markov decision process. Across all scenarios, our analysis favored releasing parrots into population 2 first when population sizes were small, followed by releases into population 3, and subsequently to population 1. The point in the decision space at which these transitions occurred was a function of location demographic rates. Releasing parrots in a location harboring an extant population versus a location that does not is advisable if differences in environmental conditions that promote population growth between locations are small. If management actions cannot ameliorate limiting factors in a location, decisions will favor translocations from such locations and allocations to locations where growth is more likely. When the cost of conducting a release is considered, managers should occasionally postpone releases to avoid these costs. This is optimal whenever the weighted contribution of the additional parrots to the sum of the discounted additional expected future population levels is less than the size of the fixed costs. This was most often the case for population 1 as costs increased. © 2013 The Wildlife Society.