Canopy spectral characteristics of apple tree are the basis for apple physiological monitoring with remote sensing and production management. This study carried out the measurement of apple tree canopy spectral reflectance at apple fruit stage in Qixia City. The reflective spectral features and sensitive wavebands of apple tree canopy at fruit stage were initially explored and regression models between sensitive bands and the ratio of fruit/tree were established with digital photography techniques. The non-destructive estimation of fruit amount (fruit/tree ratio) was realized. The results showed that the spectral features of apple tree canopy at fruit stage mainly presented as absorption of blue light and red light, reflection of green light, and the strong reflection of near-infrared between 750nm to 1300nm, with two reflection peaks near 1650nm and 2200nm. The sensitive bands to fruit amount were 435nm, 670nm, 730nm, 940nm, 1140nm and 1480nm. Vegetation indices of difference, ratio and normalized difference were built respectively with the sensitive bands, and the best spectral parameters were selected. Estimation models of fruit amount (fruit/tree ratio) were established accordingly and the best model was evaluated and validated as y = 0.0086\[NDVI (940,730)\]2-1.0934NDVI (940,730) + 0.3209. This model is expected to provide a relatively fast and accurate estimation method for canopy fruit amount at apple tree fruit stage.