Objective: In order to realize the fast and accurate prediction of moisture content of fresh fruit corn. Methods: Hyperspectral technology was used to collect and extract the spectral data of fresh fruit and corn. The effects to the accuracy of model were studied by comparing the data from Standard Normalized Variate (SNV), Multiplicative Scatter Correction (MSC), Savitzky-Golay smooth (SG) and Moving Average (MA), etc. MSC was selected for preprocessing. Based on the data preprocessed by MSC, successive projections algorithm (SPA), Competitive Adaptive Reweighted Sampling(CARS）and random frog (RF) were selected to optimize the characteristic wavelength for the prediction of moisture content of fresh fruit corn. Results: It showed that the prediction effect of moisture content of MSC-CARS-PLS model was the best. The coefficient of determination (R²_p) of prediction set was 0.825 0, the predicted error (RMSEP) was 0.006 0. Conclusion: It is showing that the rapid nondestructive testing of moisture content of fresh fruits and corn can be realized by using hyperspectral technology.