Abstract: The similarity measures between SAR images are the basis for target recognition, image registration and so on. In this paper, a new similarity measure for uncertain SAR images based on intensity increment code is proposed after analyzing the characteristics of existing methods. Firstly, the code image is obtained by averaging the difference of brightness in adjacent pixels in SAR images. Then, the similarity measure for uncertain SAR images is got by evaluating the concordance of code images. The proposed method is proved to be robust to speckle noise, partially occlusion and fuzzification through theoretical analysis in this paper and the SAR image matching at confidence level using the proposed method is discussed. The theoretical analysis and experimental results demonstrate that the proposed similarity measure is robust to the speckle noise, partially occlusion and fuzzification in SAR images and is valid for uncertain SAR image matching.