To recognize early hidden bruises on kiwifruits, taking ‘Huayou’ kiwifruits as research object and near-infrared (NIR) hyperspectral imaging system as detection device, the NIR hyperspectral image of kiwifruits with no bruises and hidden bruises damaged in 3h were collected. The sample images were covered a wavelength of 900~1700nm. It was found that the no bruises and hidden bruises kiwifruits had obvious reflectance value between 900nm and 1350nm. Principal component analysis (PCA) was conducted on this spectral region, and it was found that 1050~1200nm was the optimal band for identifying kiwifruits with hidden bruises. Based on the weight coefficients of the principal component (PC) images in the optimal band and their correlation coefficients of reflection values between wavelengths, four characteristic wavelengths, i.e. 1057, 1090, 1120 and 1177nm, were extracted. After a secondary PCA, median filtering, threshold segmentation and mathematical morphology, recognition algorithm of kiwifruits with hidden bruises were proposed and then applied on 70 no bruises and 70 hidden bruises kiwifruits. Results showed that the correct identification rates for no bruises kiwifruits and hidden bruises kiwifruits reached to 100% and 95.7%, respectively. The average identification rate was 97.9%. This study indicates that NIR hyperspectral imaging technology is effective for identifying kiwifruits with early hidden bruises. Its results provide basic data for developing online detection system for bruised kiwifruits.