Disaster management software deals with supporting staff in large catastrophic incidents such as earthquakes or floods, e.g., by providing relevant information, facilitating task and resource planning, and managing communication with all involved parties. In this paper, we introduce the SoKNOS support system, which is a functional prototype for such software using semantic technologies for various purposes. Ontologies are used for creating a mutual understanding between developers and end users from different organizations. Information sources and services are annotated with ontologies for improving the provision of the right information at the right time, and for connecting existing systems and databases to the SoKNOS system using those annotations. Furthermore, the users' actions are constantly supervised, and errors are avoided by employing ontology-based consistency checking. We show how the pervasive and holistic use of semantic technologies leads to a significant improvement of both the development and the usability of disaster management software, and present some key lessons learned from employing semantic technologies in a large-scale software project.