The time-honored principle that “a ship is its own best lifeboat” is a foundation of large passenger ship design and operation. Given the significant risks inherent in the evacuation process, reducing the likelihood that a ship will have to do so because of an emergency is clearly desirable. However, incidents occur which can or will result in the loss of the ship and evacuation is unavoidable. Delays in the evacuation process can have a catastrophic effect on the outcome, usually in terms of human lives. The foundering of the Costa Concordia after grounding and holing five watertight compartments is an illustrative example of this. It has been widely observed that had the ship not fortuitously drifted towards land before capsizing, the loss of life would likely have been significantly higher. The decision regarding whether to begin evacuation was not made until over an hour after the ship struck bottom, which complicated evacuation and rescue efforts. The first question considered in this paper is, based upon the information available to the master, was this delay in deciding to evacuate extreme or typical of such incidents. The second question to be considered is what factors drive a vessel master’s decision whether and when to initiate evacuation. Factors regarding the severity of the incident make evacuation either more or less urgent and affect the time available for a decision. Other factors provide alternative solutions or deter the prospect for a successful evacuation, which can delay or preclude a decision. The third question considered is what length of decision delay in starting preparations for evacuation is necessary or expected.