Deregulation and restructuring of electric power supply business are proceeding all over the world. In many cases, a competitive environment is introduced, where a market to transact electric power is established, and various attempts are done to decrease the price. On the other hand, environmental problems are pointed out in recent years. However, there is a possibility of the environmental deterioration by cost reduction of electric power. In this paper, the RPS (Renewable Portfolio Standard) system is taken up as the solution method of environmental problem under the deregulation of electric power supply business. A RPS model is created by multi-agent theory, where Q-learning is used as a decision-making technique of agent. By using this model, the RPS system is verified for its effectiveness and influence.