Abstract: Visible / near infrared hyperspectral imaging system ranging from 400 nm to 1000 nm was investigated for non- destructive determination of protein content,p H value and tenderness in chilled mutton. Collected the hyperspectral scattering image of chilled mutton and extracted the reflectance spectrum at range of interest.Firstly,the partial least squares regression( PLSR) model was established by the pretreatment of original spectrum to prefer the best pretreatment method. Then,the successive projections algorithm( SPA) and competitive adaptive reweighed sampling( CARS) were used to select characteristic wavelengths,the PLSR prediction models of protein content,p H value and tenderness under the characteristic wavelengths were set up and analyzed. The results showed that using original spectrum to establish OS- PLSR models for predicting protein content,p H value and tenderness were better than the PLSR models constructed by pretreatment spectrum; the PLSR prediction model of protein content under full wavelengths has the best effect,the correlation coefficient of prediction( Rp) and the predict root mean square error( RMSEP) were 0.869 and 0.097. The result of SPA- PLSR model for predicting p H value was ideal and RP value was 0.958,RMSEP value was 0.067. The result of CARS- PLSR model for predicting tenderness has high predictive power and the RP value was 0.862,RMSEP value was 0.706.Therefore,visible / near infrared hyperspectral imaging technology is feasible for non- destructive determination of protein content,p H value and tenderness in chilled mutton.