The principle of the radiation calibration of TDI-CCD spatial stereo camera is introduced in detail in three aspects of dark current data, relative calibration and absolute calibration. Then the process of the radiation calibration is briefly described. Finally the data measured during the calibration is disposed and analyzed. How the dark current is eliminated effectively and the method of the relative calibration in each condition is mainly illustrated. In order to effectively eliminate the dark current, a part of the direct current, two methods are compared. In the process of relative calibration, the calibration matrices are calculated after analyzing a huge number of data. And in the tolerance of the system, the most suitable matrix to any condition is optimized among many matrices. It is proved that the average deviation of any calibrated image is usually between 0.08 and 1.5 in gray value, and the maximum error is 2.80%. So the complexity of the calibrating process is decreased in the practical application, and the image processing in the later stage of the work is also simplified.