Abstract The world is losing its biodiversity at an alarming rate and many agencies are committing to considerable investment in global conservation. Given the enormity of environmental issues, the funding available to managers is insufficient. Managers must make decisions about how to act within the bounds of this limited funding. Conservation decision-making is also limited by a lack of knowledge about the systems we are trying to conserve. Much of the information required for effective conservation is uncertain. In this thesis I focus on practical ways of approaching the immense predicament of how to make good conservation decisions in the face of these two limitations. In chapter two I provide both an optimal framework and analytic rule of thumb for allocating limited funding among subpopulations of a threatened species. My results show that the number of subpopulations we can effectively manage is driven by the economic constraints placed on management and the risk of extinction of the species we are trying to protect. We discover that it is rarely optimal to manage all the remaining isolated subpopulations of a threatened species. This highlights the importance of a triage approach to the management of subpopulations of a threatened species under the current climate of limited funding, leading us to coin the term ‘subpopulation triage’. One key area of uncertainty that links directly with how we allocate resources for conservation is uncertainty in the impact of our management actions on the systems we are trying to protect (the impact-investment curve). This relationship often drives the outcomes of our decision-making frameworks. In chapter three I investigate how uncertainty in the impact-investment curve, assumed in chapter two, alters our optimal management decision. Again, I find that limited conservation finances are a major limiting factor in the robustness of a strategy to our incomplete understanding. I discover that ‘subpopulation triage’ can be a natural consequence of robust decision-making. Uncertainty is not, however, always beyond our control. We can reduce it by diverting funding from management to collect data on our systems. This entails monitoring costs that must also be considered when making optimal conservation decisions. There are a number of reasons why we could monitor; to reduce our uncertainty in the status of threatened species where management is driven by species status; to aid learning about a component of the system we are managing; for both initial surveillance and adaptive approaches; and to report on the performances of conservation action to stakeholders. In chapter four I assess the benefit of initial surveillance to gain information on biodiversity value before we acquire a land parcel for the reserve network. The risk here is that the land parcel may be removed from the market during surveying. I describe both an optimal method, using stochastic dynamic programming (SDP), and a simple rule of thumb, for how to make such decisions. The solutions to this problem illustrate how optimal conservation is necessarily dynamic and that immediate implementation of a conservation plan may not always yield the best conservation outcome. Learning does not always need to take place in the absence of management. In chapter five I investigate adaptive learning for a threatened species where we must discriminate between multiple hypotheses of how the system works by implementing different management actions. We find that the optimal action depends on our belief in each model being the true model of our system, the benefit from each action under each model, and the number of sites available to implement an active adaptive strategy. In chapter six I investigate when one should learn about the state of the system through monitoring when management is state-dependent. Here our management of subpopulations of a threatened species is based on whether these subpopulations persist. I ask when should we survey or manage a subpopulation, and when, if ever, should we do nothing in a subpopulation of a threatened species. I find that management actions should not only be driven by the return on investment gained by managing a subpopulation but also by our certainty of the persistence of a subpopulation. This is the first work to show a direct trade-off between return on investment from conservation action and reduced uncertainty. One key evaluation method currently adopted worldwide is the use of ‘State of the Environment’ reporting. In chapter seven I assess the flaws of ‘State of the Environment’ reporting, the current method adopted worldwide for evaluating conservation policy. I show the positive biases inherent in such reporting and provide a new metric for reporting on conservation performance that is simple, transparent and provides an unbiased report on performance in reaching conservation objectives. I show that without honest reporting of conservation gains – and losses – we limit our ability to assess where we are in terms of conservation progress. Overall my thesis shows the need for managers to consider a triage approach to threatened species management, not as a process of giving up, but as a tool for ensuring species persistence in light of the urgency of most conservation requirements and the realities of financial and knowledge limitations. Indeed if conservation is a field dedicated to the protection of biodiversity then those responsible for decision-making––politicians, scientists and environmental managers––must use whatever approach gives the best outcome for the environment. Under current limitations, triage is often a necessity not an option.