The visible and near infrared spectral reflectance was measured for the leaves by using Shimadzu UV—2450 spectrograph. At the same time, the nitrogen content of apple tree leaves were measured in Kjeldahl nitrogen method. The rapid determination model of apple leaf nitrogen content suitable for portable detector was built. The spectral properties were explored and the analysis of correlation was conducted among the nitrogen content of apple tree leaves and their spectral reflectance. Then two sensitive wavebands (652nm, 772nm) were obtained. At the same time, by adopting the method of segment reduced precise sampling, the detailed NDVI within the range of 350~730nm and 740~880nm were constructed, and then the wavebands (859nm,364nm) of the highest correlation with nitrogen content were selected to construct NDVI for apple tree. Finally the nitrogen partial least square regression model of apple tree leaves was established based on apple tree NDVI (859nm,364nm) and reflectance at 652nm and 772nm. The R2 of the calibration model reached to 0.9048, RMSEC was 0.1597, and the R2 of predicted model reached to 0.917, RMSEP was 0.2833. The model reaches practical level and can be used as the nitrogen content rapid determination model to develop a portable apple tree nutrition detector.