Herbal tea, a kind of traditional Chinese beverage, is widely used in China for its health benefits. This kind of tea beverage usually consists of several traditional Chinese medicines. So the taste of different herbal tea species varies a lot for their diverse components. In this paper, herbal tea discrimination method based on electronic nose (EN) is investigated. A laboratory developed EN, consisting of eight metal oxide semiconductor (MOS) gas sensors, is used in this study. EN responses to herbal tea samples at three temperatures’ environment (20, 30, and 50 °C) are collected by a desktop. Principal component analysis (PCA) and stochastic resonance (SR) is used for EN measurement data processing. PCA results indicate that herbal tea samples are discriminated more clearly with the increase of measurement temperature. The SR output signal-to-noise ratio (SNR) parameter is used for herbal tea discrimination. PCA method can not discriminate all herbal tea samples under different temperature. SR SNR method successfully discriminates all herbal tea samples under 20, 30, and 50 °C. Denglao herbal tea SNR eigenvalues have much difference from other tea samples. Higher measurement temperature leads to better herbal tea discrimination based on EN.