Marking animals so that they are uniquely identifiable provides information that may assist conservation efforts. Nevertheless, some methods used to mark animals can be harmful. We used mathematical methods to assess the trade-off between the impact of marking threatened species and the value of the information gained. We considered the case where 2 management strategies, each aiming to improve a species' survival rate, are implemented in an experimental phase. The results of the experiment were applied in a postexperimental management phase. We expressed the expected number of survivors in both phases mathematically, accounting for any mortality caused by the experiment, and determined the proportion of animals to mark to maximize this number. The optimal number of animals to mark increased with the number of individuals available for the experiment and with the number of individuals to be managed in the future. The optimal solution was to mark only 25% of the animals when there were 1000 individuals available for the experiment, the results were used to manage 2000 individuals, and marking caused mortality of 1%. Fewer animals were marked when there were fewer animals in either phase or when marking caused higher mortality. In the case of the Helmeted Honeyeater (Lichenostomus melanops cassidix), the optimal proportion to mark was <1 if the mortality rate was >0.15%-1%, with the threshold depending on the number of animals in the experimental and postexperimental phases. The trade-off between gaining more information about a species and possibly harming individuals of that species by marking them is difficult to assess subjectively. We show how to determine objectively the optimal proportion of animals to mark to enhance the management of threatened species.