Classical negotiation models are based on a centralised decision making approach which assumes the availability of complete information about negotiators and unlimited computational resources. These negotiation mechanisms are ineffective for supporting real-world negotiations. This paper illustrates an agent-based distributive negotiation mechanism where each agent's decision making model is independent to each other and is underpinned by an effective evolutionary learning algorithm to deal with complex and dynamic negotiation environments. Initial experimental results show that the proposed genetic algorithm (GA) based adaptive negotiation mechanism outperforms a theoretically optimal negotiation mechanism in environments constrained by limited computational resources and tough deadlines. Our research work opens the door to the development of practical negotiation systems for real-world applications.