E-commerce is considered a key service within modern information society, and the idea of automating e-commerce transactions has attracted much interest in recent years. A multi-agent model is a system that applies various autonomous agents to accomplish specified goals. Such a system addresses resource allocation issues. Because the nature of resource trading requires multiple agents to request geographically dispersed heterogeneous resources, we use a multi-agent architecture for e-commerce because each agent can be describe each participant intelligently. In this paper, negotiation agents based on fuzzy ECA rule-based proposed. Here we focus on agents in e-commerce that negotiate between sellers and buyers in order to get the best deal. The negotiation process between buyers and its sellers begins through combined and fairness protocols. We add learning properties to agents based on a fuzzy decision tree to develop negotiation skills and present the results. Using a fuzzy decision tree helps us understand and adapt other agents’ behavior and real-time world conditions in order to produce the best contracts. Thus, the agent can improve in terms of skills on negotiation by updating its fuzzy decision tree.