Abstract: The production and consumption of drinks play a very important role in the economic development and the quality identification of drinks gains a great concern of the society. Therefore, it is quite necessary to develop fast and accurate methods and apparatus for the quality identification. On the basis of expounding existing methods of quality identification of drinks, this paper proposes the selecting rule of gas sensors and the constituting method of sensor array and introduces the artificial olfactory system for identifying drink quality. A new drink quality identifying system based on artificial olfactory technology is established, at the same time, a kind of RBF neural network as pattern recognition technology is adopted for the system. The experimental results show clearly that the system can be used to identify preferably both drink classes and quality grades of the same kind drink. Meanwhile, the reasonability and feasibility of constituting method of the system are proved, and some technological foundation has been laid for the system to be adopted for further commercial utilization.