A practical physics-based regression method was developed and evaluated for nearly real time estimate of land surface emissivity spectra in 8-14 μm from hyperspectral thermal infrared data. Two spectral emissivity libraries and one atmospheric profile database fully covering all the possible situations for clear sky conditions were elaborately selected to simulate the radiances at the top of the atmosphere (TOA). The regression coefficients were determined by the main principal components of emissivity spectra and those of simulated brightness temperature at TOA using a ridge regression method. The experience with the simulated Interferometer Atmospheric Sounding Instrument (IASI) data showed that the emissivity spectra could be retrieved under clear sky conditions with root mean square errors of 0.015 and 0.03 for 714-970 cm(-1) (10.3-14.0 μm) and 970-1250 cm(-1) (8.0-10.3 μm), respectively, for various land surface and atmospheric conditions. This indicates the proposed method may be robust and applicable for all hyperspectral infrared sensors.