This paper deals with a criterion to reduce excessive communication in the Cellular Robotic System (CEBOT). While communication plays a significant role in acquiring necessary information for the decision making, we should consider its cost especially in the case when a large number of agents is involved. Therefore, we have been trying to achieve the “intelligent communication” that is defined as accumulation of knowledge enables higher performance and a lower frequency of robot communication. Acquiring information through communication, a probability model is constructed. The decision is based on the model but the risk for that is evaluated with respect to whether communication is negligible. Considering both communication cost and risk for decision making that is derived from the game theory approach, we will set a criterion for communication necessity and show that there exists a correlation between accuracy of the estimate model and communication frequency and estimated risk through the simulation results.