Abstract: Objective: To establish a rapid detection method of consistency evaluation and astilbin content as an index component for Smilax glabra formula granules. Methods: Based on portable near infrared spectrometer, the consistency evaluation model was established by using software OMNIC, and a quantitative model of astilbin was established by combining partial least squares (PLS) and support vector regression (SVR). Results: The fingerprint characteristics of 55 batches of Smilax glabra formula granules were similar, and the similarity was 0.9928-0.9996. The PLS and SVR quantitative models of astilbin showed that the correlation coefficient (R_P) of prediction set was 0.8483 and 0.8990, respectively, and the root mean square error of prediction (RMSEP) was 1.809 and 1.481, respectively. Conclusion: Based on portable near infrared spectrometer, a rapid evaluation method of Smilax glabra formula granules is established in this study. The established model can predict the content of index components quickly and non-destructively, which provides some reference for the quality control of Smilax glabra formula granules.