Evaluating complex emergency operations has in the past been an elusive target for evaluators. The speed of operations, the magnitude of the support mobilized and the changing character of interventions over time – initial high intensity of humanitarian operations followed by rehabilitation programmes, with a longer-term perspective of reconstruction and development – requires a flexible approach that is difficult to incorporate in orthodox (usually mid-term or final) evaluations. In recent years, a new set of “evaluative exercises” has emerged, among them the After Action Review (AAR), the Real Time Evaluation (RTE), the Iterative Evaluations with Mini-seminars (IEMS), etc. Each of these exercises has two features in common: -A closer link between implementation and learning, by having an evaluation before the end of the project so that some recommendations can be rapidly translated into action; and -Capturing the emergency while operations are ongoing, while most of the actors still have a fresh memory.