A method of ambient light correction was proposed to evaluate the sugar content of Huanghua pears on tree by visible /near infrared diffuse reflectance spectroscopy (Vis/NIRS). Due to strong interference of ambient light, it was difficult to collect the efficient spectral of pears on tree. In the field, covering the fruits with a bag blocking ambient light can get better results, but the efficiency is fairly low, the instrument corrections of dark and reference spectra may help to reduce the error of the model, however, the interference of the ambient light cannot be eliminated effectively. In order to reduce the effect of ambient light, a shutter was attached to the front of probe. When opening shutter, the spot spectrum were obtained, on which instrument light and ambient light acted at the same time. While closing shutter, background spectra were obtained, on which only ambient light acted, then the ambient light spectra was subtracted from spot spectra. Prediction models were built using data on tree (before and after ambient light correction) and after harvesting by partial least square (PLS). The results of the correlation coefficient(R) are 0.1, 0.69, 0.924; the root mean square error of prediction(SEP) are 0.89°Brix, 0.42°Brix, 0.27°Brix; ratio of standard deviation(SD) to SEP(RPD) are 0.79, 1.69, 2.58, respectively. The results indicate that, method of background correction used in the experimentcan reduce the effect of ambient lighting on spectral acquisition of Huanghua pears in field, efficiently. This method can be used to collect the visible/near infrared spectrum of fruits in field, and may give full play to visible/near-infrared spectroscopy in preharvest management and maturity testing of fruits in the field.