Effective reuse of parts is necessary to achieve a sustainable society. For that purpose, individual parts should be used with little environmental load. Moreover, because of variety of consumers' preferences and usages, parts should be used with consideration of consumer's characteristics as well as the information on parts. We have developed the part agent system using network agents and RFID technology to manage parts throughout their life cycle. In previous researches, functions of part agents have been developed that supports recall of parts and replacement of parts in consideration of consumers' preference. However, it is also required for part agents to accommodate the state of parts and consumers' characteristics. For that purpose, we developed a function to propose maintenance activities of part using Bayesian Network model. The function calculates the probability of uncertain phenomenon on the part using the model based on the situation surrounding the part such as sensory information and choice of consumers.