The traditional chemical-physical analyzing method for beverage inspection is troublesome and inefficient. Recently, a novel electronic nose system was developed, which can inspect the odorant from beverage quickly and accurately. This system is mainly composed of a gas sensor array and a data processing apparatus. In order to minimize the effect of inspecting environment, nitrogen was used as carrier gas, which could carry the odorant from the sample vessel into the sensor chamber. Four feature parameters were picked up from each sensor reacting curve to improve the rate of Signal-to-Noise. Moreover, the principal component analysis (PCA) and the back-propagation(BP) neural network were used to identify beverage samples. The results demonstrate that this inspecting method is efficient and the recognition rate is up to 95.2%.