The invention provides an anesthesia depth monitoring system and method based on near infrared spectroscopy. The anesthesia depth monitoring system comprises a cerebral blood oxygen signal acquisitionmodule A, a sample entropy calculating module B, a target signal identification module C and an optimal threshold determining module D; first, various Hb and HbO2 signals of the prefrontal lobe of apatient are collected through near infrared light with the wavelength being 735 nm and near infrared light with the wavelength being 850 nm correspondingly, and band-passing filtering of 0.01-0.04 Hzis conducted; then data processing is conducted by a sample entropy algorithm, and the sample entropy values of the patient at different stages are calculated; by drawing an ROC curve, signals with the highest distinguishing capacity to an anesthesia state and a waking state are selected by an ACU value; at last, the optimal threshold value capable of distinguishing the anesthesia stage and the waking stage is found through a youden index, and the anesthesia depth is monitored by the threshold value. The anesthesia depth monitoring system and method have the advantages that the data length required by calculating is small, the operation efficiency is high, the signals needing to be monitored can be screened out more quickly, safety and convenience are achieved, and the prince is low.