Abstract. Ambient intelligent systems are defined as being able to perceive their environment, being aware of the presence of people and other agents, and respond intelligently to these agents’ needs. Today the hardware requirements for achieving these capabilities are met. Earlier work have argued that knowledge intensive case-based reasoning is a feasible method for ambient intelligence. In this paper that argument is supported by testing of an implementation in a hospital ward domain, which shows that despite some issues related to the current implementation the case-based reasoner performs at an acceptable level.