Near infrared spectroscopy data from 135 apple samples of different storage periods were determined, the charactar of them was extracted and analyzed using principal components analysis. Therefor an ANN model for detection of apple chewiness was established. Our results showed that the preprocessing of spectrum scattering was the weighted multiple scatter correction(WMSC) and mathematics processing was “2441”. The structure of the artificial neural network mode was 3—16—1, established after extracting 3 principle component as the characteristic variables of the original information. The decision coefficient of our model on validation is 0.992 4, and the root mean square error is 0.000 108 2. Our results confirmed that the near infrared spectroscopy technology can use to detect the chewiness of apple rapidly, without forecast destructive.