1Institute of Vegetables and Flowers，State Key Laboratory of Vegetable Biobreeding，Chinese Academy of Agricultural Sciences，Key Laboratory of Vegetables Quality and Safety Control，Ministry of Agriculture and Rural Affairs，Quality and Safety Risk Assessment Laboratory of Vegetable Products，Ministry of Agriculture and Rural Affairs，Beijing 100081，China；2Beijing Center for Quality and Safety of Agricultural Products，Beijing 100120，China
Abstract：The purpose of this study was to use near-infrared reflectance spectroscopy（NIRS）to
establish a near-infrared prediction model of 6 types in tomatoes and to provide the theoretical basis．377 fresh tomato samples were selected as the calibration sets and 94 were used as the validations sets．NIRS used with combination of modifi ed partial least squares（MPLS）to construct and verify the pros and cons of establishing a prediction model．The results showed that the cross-validation correlation coeffi cients（1-VR）of the tomato dry matter，total acid，soluble sugar，Vitamin C and soluble solid content were 0.979，0.851，0.828，0.832 and 0.959，all exceeding 0.80，which can be used for actual prediction．The content of lycopene prediction models 1-VR were 0.767，lower than 0.80．The correlation of the lycopene constructed models was poor，and the models need further optimization．
