Limb imaging spectrometer is an important new remote sensor for research and application. Signal-to-noise ratio (SNR) is one of the key parameters to quantitatively evaluate the image quality and radiometric performance of an imaging spectrometer. The estimation and testing of SNR are very important for developing an imaging spectrometer. From the perspectives of radiative transmission and energy conversion, the SNR model is proposed, and the SNR equation of dispersive-type limb imaging spectrometer is derived, and the SNR values under several observing conditions for an limb imaging spectrometer prototype developed are theoretically evaluated based on atmospheric radiative transfer code MODTRAN 4.0. The results show that the SNR of the prototype under typical viewing geometry is not less than 8. As experimental validation, SNR testing was performed using an internally illuminated integrating sphere, and the experimental results have proved the correctness of this theoretical model.