A rapid technique is necessary to detect the natural gas which is a more and more significant fuel resource in modern industry. Terahertz (THz) technique was employed in this research to detect the principal hydrocarbon components of natural gas including methane, ethane, and propane. Two- and three-component mixtures were measured by THz setup, respectively. The amplitude ratio and time delay deviation of THz peaks between samples and reference were calculated. Phase projection pictures were obtained between the component concentrations and the amplitude ratio as well as time deviation. The phase figures evidently reflected the concentration dependent THz response, and a greatly different distribution was located in the whole phase projection area. In addition, back-propagation artificial neural networks method was utilized for the quantitative determination of components concentration and total pressure, and the correlation coefficient of the prediction set was proved to be 0.9859. Therefore, THz technique can satisfy the increasing need of rapid and efficient detection in the natural gas industry.