Abstract: Objective:In order to meet the requirements of breeding, cultivation and agricultural practice of high protein quinoa, and determine a quickly and nondestructive measurement of quinoa grain crude protein content method. Method:In this study, the rapid detection of the crude protein content in quinoa grains was systematically studied by using near-infrared spectroscopy. 122 representative quinoa varieties were selected as the test materials, among which 94 were used as the modeling set and 28 were used as the verification set. The Unscrambler 10.4 software was used to preprocess the original near-infrared spectra data after scanning, and established the quantitative prediction model of quinoa kernel crude protein content by partial least square method (PLS). Results:Combing the filter fitting method (savitzky-golay, SG) and Standard Normal Variate (SNV), obtained the best result, with the model predictive value determination coefficient (R²) of 0.9380, and the component concentration analysis error (RMSE) of 0.4823.The correlation analysis of the 28 samples in the verification set showed that the determination coefficient between the predicted value and the measured value of the national standard method was 0.9416. One-way anova showed that there was no significant difference between the measured value of GB method and the predicted value of the model (P>0.05), and indicating that the established model had high accuracy and good prediction effect. Conclusion:As a simple, this method can be used as a rapid and non-destructive method for the detection of crude protein content of the quinoa seeds, and can provide technical support for the breeding, cultivation and agricultural practice of high-quality quinoa.