Leaf moisture is an important parameter of plant health, and Terahertz is sensitive to leaf moisture. When terahertz spectra penetrate the leaf, the leaf can be imaged according to different moisture contents at different positions. In this research, Terahertz technology was used to detect the isolated leaf moisture. The relationship between leaf moisture and terahertz spectroscopy data was studied. Terahertz data were obtained by scanning the leaf in vitro. Then the time domain amplitude was obtained and the frequency domain spectra were extracted to rebuild the leaf images. The distribution of the leaf moisture could be clearly observed. This study showed that the image quality with the minimum time domain performed the best. Also, the optimal model was established. The correlation coefficient and the root mean square error of the prediction model were 0.981 7 and 0.023 7 g, respectively. The result of the research provides a guidance to predict live leaf moisture using Terahertz technology in the future.