An effective approach for identification of the origin of Scutellaria baicalensis, an essential member of the family of Chinese herbal medicine and known to be an effective anti-inflammatory, is proposed based on terahertz time-domain spectroscopy (THz-TDS) and pattern recognition. Terahertz absorption spectra of Scutellaria baicalensis collected from its main growth areas in China, including Inner Mongolia, Shanxi and Shaanxi are investigated using the proposed method, in the frequency range from 0.2 to 1.7 THz. To reduce the dimensionality of the original spectral data and extract useful features of the data, principal component analysis is employed. The matrix of the selected principal component scores is fed into a classification model established by support vector machines. We use the particle swarm optimization to optimize the parameters of the classification model to achieve an identification rate of 95.56% for the samples, demonstrating that terahertz time-domain spectroscopy combined with particle swarm-support vector machines approach can be efficiently utilized for automatic identification of the origin of Scutellaria baicalensis.