Emergency management has drawn the attention of governments worldwide. It is an emerging realm for governments and organizations to respond to sudden catastrophic events rapidly and effectively during a crisis, e.g. social unrest, natural disaster, unforeseen event, and so on. There is a lot of information and knowledge involved in emergency management. The key question is how to search for the appropriate information and knowledge quickly and accurately from the massive information and knowledge based on the nature, characteristics, status and situations of the unexpected emergency in order to support the intelligent decision-making process in handling emergencies. In this paper, an unstructured information management system (UIMS) is presented for emergency management. A concept relationship model CRM) and a dynamic knowledge flow model (DKFM) of the emergency incidence have been built to organize and represent emergency knowledge. The models can support decision-makers to have better understanding of the dependence and degree of correlation between different concepts about the emergency in order to make the appropriate decisions. The system is trial implemented in a city emergency management system and its performance is evaluated and discussed