Abstract: Ten different varieties of fresh grapes, 3 different mature period and 1 kind of diseased fruit with a total of 188 samples were selected as the research object, and the near infrared spectra of these samples collected by near-infrared spectrometers. Using partial least squares (PLS), the quantitative analysis model with higher credibility by determining the content of total phenol, total sugar, fructose, sucrose and soluble solids. The correlation coefficients of indicators were above 0.90 except minority were between 0.77 and 0.89, and root mean square error were all between 0.022 and 1.410. By combining with principal component analysis in 4 199.20~9 881.46 cm^-1 spectral region, the discriminant analysis (DA) model with correctness of 92.11%, 88.89% and 96.15% for variety identification, maturity identification and diseased fruit identification respectively has been established. This study showed that near infrared detection technology can not only be used for quantitative analysis of 5 main inclusions in fresh grape, but also be used for identification of varieties, maturity and diseased fruit of fresh grape.

Key words: fresh grape, nondestructive detection, spectroscopic analysis, NIR, chemometrics