Abstract: After China's entry into the WTO, it becomes more and more urgent to study fast and precise agricultural product quality detection technology and method based on Near Infrared Spectroscopy Analysis(NIR). A NIR software system for agricultural product quality detection is developed to meet the need for quantitative software to predict constituent concentration. Some software implementation technology is adopted, such as object-oriented programming technology, vector-matrix encapsulation technology, structured storage and compound file technology, and so on. This system is composed of five functional modules, spectrum file management, spectrum displaying, spectrum pre-processing, calibration model building and management, unknown sample constituent concentration predicting, and so on. In this system, many spectral formats, such as SPC, TXT, JWS, MAT, ASF, are compatible, function and algorithms, such as PCR, PLS, GA, ANN, Wavelet, Wavelet Package, are provided. The system is of stability and security, and can be maintained and used easily. The case analysis shows that, the prediction results of PLS-NIR model and PCA-MBP-NIR model of this system are all very precise. The system is very useful, and provides very good technological support for building NIR model and concentration prediction in agricultural product quality detection.