Codebook of conventional VQ cannot be generally used and needs real time onboard updating, which is hard to implement in spaceborne SAR system. In order to solve this problem, this paper analyses the characteristic of space-borne SAR raw data firstly, and then utilizes the distortion function of multidimensional space as criterion, and finally the adaptive code book design algorithm is proposed according to the joint probability density function of the input data. Besides, the feasibility of the new algorithm in cascade with entropy coding and the robustness of the algorithm when error occurs during transmission are analysed based on the encoding and decoding scheme. Experimental results of real data show that codebook deriving from the new algorithm can be generally used and designed off-line, which makes VQ a practical algorithm for space-borne SAR raw data compression.