German Red Cross mobile hospitals must be operational in areas of war or disaster within a few days. In the ARDOR project, we collaborate with the German Red Cross in order to accelerate the work ows in these hospitals by developing and deploying a con gurationless distributed information technology on top of a mobile ad hoc network [Cooperation Agreement of the German Red Cross with the Department of Telematics of the University of Freiburg and Requirements Specification for the Use of Pervasive Computing in Mobile Hospitals]. The main applications are scheduling of resources and their allocation.The computer-supported scheduling must enable the planning sta to query the currently available resources and services and the respective booking information.To achieve this, we introduce service discovery on the middleware layer. However, service discovery has to handle higher order (medical or logistic) services, as an entry in the schedule typically consists of the simultaneous allocation of many resources. Furthermore, service discovery must be scalable as the mobile ad hoc network might be highly dynamic and large in size. This paper claims that our self-organizing and scalable service discovery approach provides a way to enable reliable access to higher order services.