The application of artificial neutral network to data processing of multi-spectral radiation thermometry was presented. By taking advantages of neutral network and various emissivity samples, the emissivity models of the targets were identified and the true temperature and spectral emissivity were available simultaneously. The measurement accuracy was improved further by subdivision. The effects of measurement errors on the measurement accuracy of temperature and emissivity were also analyzed. Computer simulation results proved that the method is an effective way for both temperature and emissivity measurements.