Abstract: Airports are crucial infrastructures in both civil and military transportation systems. Automatic detection of airports then has been attracting significant research interests in the remote sensing areas. A method for detecting airport runways in a Polarimetric Synthetic Aperture Radar(PolSAR) image using difference degree iteration is proposed in this paper. Superpixels are generated by Simple Linear Iterative Clustering(SLIC) algorithm which is applied to the PolSAR image in the segmentation step, followed by a coarse classification according to the three-component decomposition feature and scattering entropy. The fine classification, combining the enhanced K-means clustering method with the difference degree iteration, is then applied. The extraction of runways and the complete airport areas from the classification results can then be achieved by exploiting both of the scattering and geometric features. The UAVSAR data set is used for verifying the performance of the proposed method. And the results suggest that the airport runway can be detected effectively by the proposed method, while maintaining the detailed structures intact with clear edges. A low false alarm rate is also exhibited during the numerical experiments.