The utility model discloses a control method and device for the image code rate in order to realize that a proper quantification parameter is chosen for each image block, which not only can reduces the difference between the actual code rate and the demanded code rate of each image after the code rate conversion, but also decreases the accumulative error. The method comprises that the target bitsof the current image block is determined first, and the target bits refer to the target bits when the current image block output the code rate according to the requirements; then a quantification parameter is chosen which makes the actual bits of the current image block most close to the target bits among all the quantification parameters of the accumulative bits errors in all the converted imageblocks; at last the current image is quantified according to the determined quantification parameter.