[Objective] The identification of easily confused tree species in mountain area is always the focus and difficulty of remote sensing. Sensitive spectral index was introduced into cloud model to identify easily confused tree species of mountainous area, so as to improve the accuracy of tree species identification. [method] Based on ZY-3 multispectral remote sensing image, the sensitive bands and sensitive spectral indices of the species were selected by correlation analysis firstly. then the cloud model of species identification were constructed and the species were identified by maximum determination method lastly. Meanwhile, support vector machine (SVM) was applied to the same data to be a contrast. [results] The results showed that the 3 and 4 bands of ZY-3 image are more sensitive. Based on the sensitive spectral indices, the recognition accuracy of cloud models: Quercus acutissima was 91.65%, Robinia pseudoacacia was 89.49%, the overall accuracy was 90.80%; the recognition accuracy of SVM: Quercus acutissima was 85.94%, Robinia pseudoacacia was 83.78%, the overall accuracy was 84.86%. Based on the sensitive bands, the recognition accuracy of cloud models: Quercus acutissima was 86.20%, Robinia pseudoacacia was 80.33%, the overall accuracy was 83.27%; the recognition accuracy of SVM: Quercus acutissima was 83.82%, Robinia pseudoacacia was 68.26%, the overall accuracy was 76.04%. [Conclusion] Compared with the SVM, cloud model constructed by both sensitive band and sensitive spectral index can effectively improve the recognition accuracy of tree species. Compared with sensitive band, both cloud model and SVM constructed by sensitive spectral index can effectively improve the identification accuracy of tree species, especially for Robinia pseudoacacia. In the inversion and mapping of the tree species, cloud model can be used to eliminate the influence of complex vegetation outside the target vegetation by the membership degree calculation, which provides a new technical support for the identification of easily confused tree species and remote sensing mapping in complex mountainous areas.[Objective] The identification of easily confused tree species in mountain area is always the focus and difficulty of remote sensing. Sensitive spectral index was introduced into cloud model to identify easily confused tree species of mountainous area, so as to improve the accuracy of tree species identification. [method] Based on ZY-3 multispectral remote sensing image, the sensitive bands and sensitive spectral indices of the species were selected by correlation analysis firstly. then the cloud model of species identification were constructed and the species were identified by maximum determination method lastly. Meanwhile, support vector machine (SVM) was applied to the same data to be a contrast. [results] The results showed that the 3 and 4 bands of ZY-3 image are more sensitive. Based on the sensitive spectral indices, the recognition accuracy of cloud models: Quercus acutissima was 91.65%, Robinia pseudoacacia was 89.49%, the overall accuracy was 90.80%; the recognition accuracy of SVM: Quercus acutissima was 85.94%, Robinia pseudoacacia was 83.78%, the overall accuracy was 84.86%. Based on the sensitive bands, the recognition accuracy of cloud models: Quercus acutissima was 86.20%, Robinia pseudoacacia was 80.33%, the overall accuracy was 83.27%; the recognition accuracy of SVM: Quercus acutissima was 83.82%, Robinia pseudoacacia was 68.26%, the overall accuracy was 76.04%. [Conclusion] Compared with the SVM, cloud model constructed by both sensitive band and sensitive spectral index can effectively improve the recognition accuracy of tree species. Compared with sensitive band, both cloud model and SVM constructed by sensitive spectral index can effectively improve the identification accuracy of tree species, especially for Robinia pseudoacacia. In the inversion and mapping of the tree species, cloud model can be used to eliminate the influence of complex vegetation outside the target vegetation by the membership degree calculation, which provides a new technical support for the identification of easily confused tree species and remote sensing mapping in complex mountainous areas.