Ex-situ conservation strategies for threatened species often require long-term commitment and financial investment to achieve management objectives. Here, we interpret the decision to adopt ex-situ management for a target species as the end point of several linked decisions. Logically, one must first decide which specific management actions are most likely to achieve the fundamental objectives of the recovery plan, with or without the use of ex-situ populations. Once this first decision has been made, one can decide whether to establish an ex-situ population, accounting for the probability of success in the initial phase, for example the probability of successful breeding in captivity. Approaching these decisions in the reverse order (attempting to establish an ex-situ population before its purpose is clearly defined) can lead to a poor allocation of resources. We use the recovery program for the threatened spotted tree frog (Litoria spenceri) in south-eastern Australia as an example to illustrate our decision framework. Across a range of possible management actions, only those including ex-situ management were expected to provide > 50% probability of species' persistence, but they came at a greater financial cost than in-situ-only alternatives. The expected benefits of ex-situ actions would also be offset by additional uncertainty and stochasticity associated with establishing and maintaining ex-situ populations. Naïvely implementing ex-situ conservation strategies can lead to inefficient management. We provide a framework to help managers explicitly evaluate objectives, management options and the probability of success prior to establishing a captive colony of any given species. This article is protected by copyright. All rights reserved.