Decision theory provides an organised approach to decision making in natural resource conservation. The theory requires clearly stated objectives, decision alternatives and decision-outcome utilities, and thus allows for the separation of values (conservation and other societal objectives) from beliefs. Models express belief in the likely response of the system to conservation actions, and can range from simple, graphical representations to complex computer models. Models can be used to make predictions about likely decision-outcomes, and hence guide decision making. Decision making must account for uncertainty, which can be reduced but never eliminated. Uncertainty can be described via probabilities, which in turn can be used to compute the expected value of alternative decisions, averaging over all relevant sources of uncertainty. Reduction of uncertainty, where possible, improves decision making. Adaptive management involves the reduction of uncertainty via prediction under two or more alternative, structural models, comparison of model predictions to monitoring, and feedback via Bayes’ Theorem into revising model weights, which in turn influences decision making. As part of a 3-day workshop on structured decision making (SDM) and adaptive resource management (ARM), we constructed a prototypical decision model for the recovery for Hector’s dolphin (Cephalorynchus hectori), an endangered dolphin endemic to New Zealand coastal waters. Our model captures several steps in the process of building an SDM/ARM framework, and should be useful for managers wishing to apply these principles to dolphin conservation or other resources problems.