Effective manipulation of context is very important in providing the context-aware services. In recent years, a variety of context models have been proposed to p roperly handle the key aspects of the context, while focusing on scenario-based acquisition, management, and representation of context. However, they are difficult to be employed for the agent-based system requiring distributed reasoning. In this paper we propose a context modeling approach for distributed reasoning and merge operator reflecting the perspective of constituent agents for rational reasoning. In addition, an agent-based context-aware system is developed implementing the proposed scheme. Performance evaluation by computer simulation on a use case of smart classroom shows that the proposed approach enables the agents to rationally reason and thereby provide intelligent context -aware services to the users.