The impact of humanitarian assistance is a key question for all of the stakeholders involved, including donors, aid agencies and the affected population. It would clearly be helpful, in deciding what sort of assistance to provide or how much is needed, to be able to answer questions about the difference aid made, or how many lives were saved. However, current practice in this area is generally poor; bold claims are often made about the impact of humanitarian assistance on the basis of limited evidence. This paper examines how the impact of humanitarian aid can be measured, why this is increasingly being demanded and whether it is possible to do it better.