Flavonoid is an important index affecting the nutritional value of peanut seed. The spectrophotometry and chromatography to detect flavonoid content are time-consuming and labor-intensive,so they are not suitable for mass detection in breeding process,and constructing its near-infrared measurement model can provide an important technical guarantee for rapid detection of flavonoid content in peanut seed. This study used 290 peanut germplasms with different flavonoid content to construct the model. The flavonoid content determined by the Al3+ chromogenic method was between 46.96 and 140.18 mg RT（RT: rutin）·(100 g)-1. The near-infrared spectrum of peanut seed was scanned and collected by Perten DA7250 near-infrared analyzer (950～1 650 nm). The partial least squares regression (PLSR) in the full wavelength spectrum range was used to compare the single and compound pretreatment methods,and the correlation coefficients and errors of different models were compared to predict the best model. The best spectral pretreatment method to determine the NIR calibration model of flavonoid content was “Savitzky-Golay derivative+baseline+de-trending”. The correlation coefficient (Rc) of calibration set was 0.884,and the root mean square error (RMSEC) of correction was 4.998. 50 peanut samples verified the model. The predicted correlation coefficient Rp was 0.904,while the predicted RMSEP was 1.122. The near-infrared spectroscopy model constructed in this study can be used to determine the content of flavonoids in peanut seeds non-destructively and efficiently. It could be effectively used to breed peanut varieties with high flavonoid content and can help construct near-infrared spectroscopy models of substances with low content (μg·g-1).