People have always sought to avoid being hit by a hazard. Whenever this has happened, however, they have tried to protect themselves by limiting damage to prevent disaster. Whatever the nature of people’s protective measures, these have always been based on their understanding and interpretation of hazards and disasters. Modern science claims to analyze hazards and disasters in detail by articulating them in ways that go beyond a local or even national context (e.g., climate change). Against the background of such an understanding it is assumed that communities can learn better from disasters and, in this way, develop more efficient mitigation measures. Here we will focus on floods that caused significant damage and were, therefore, seen as disasters or near disasters. We define learning simply as a process in which knowledge about disasters, their emergence, dynamics and outcomes is brought to bear on the mitigation of the impact of hazards. This can be done through early warning, technical measures of protection, and improved ways of coping or risk avoidance. However, learning in the sense of understanding and interpreting as such is not enough; the mitigation of hazards requires that (new) insights are applied in ways that are relevant to decision making. The successful translation of this learning into action rests on the way in which the processes, the effectiveness of the measures taken by the institutions responsible for mitigation and the compliance of the people affected by the hazards are understood. While our immediate interest falls clearly on the latter point, our account of the process of learning takes on board more recent definitions of disaster that focus on time and space (cf. Hitzler et al. 2012), and thereby emphasize the dynamic nature of underlying processes as well as the localization of events.