The paper firstly points out the defect of conventional temperature and emissivity separation algorithms when dealing with hyperspectral FTIR data: the conventional temperature and emissivity algorithms can not reproduce correct emissivity value when the difference of ground-leaving radiance and object's blackbody radiation at its true temperature and the instrument random noise are on the same order, and this phenomenon is very prone to occur in the extremity of 714-1250 cm-1 in the field measurements. In order to settle this defect, a three-layer perceptron neural network has been introduced into the simultaneous inversion of temperature and emissivity from hyperspectral FTIR data. The soil emissivity spectra from the ASTER spectral library have been used to produce the training dataset, and the soil emissivity spectra from the MODIS spectral library have been used to produce the test dataset, the result of network test shows the MLP is robust. Meanwhile, ISSTES algorithm has also been used to retrieve the temperature and emissivity from the test dataset. By Comparison the result of MLP and ISSTES, we find MLP can overcome the disadvantage of conventional temperature and emissivity separation algorithms, although the RMSE of derived emissivity using MLP is lower than ISSTES as a whole. Hence, the MLP can be regarded as a beneficial complementarity to the conventional temperature and emissivity separation algorithms.