In order to improve the accuracy and the identification speed of iris recognition, an iris recognition method is presented based on regional grayscale changes. Firstly，the normalized iris images are obtained by preprocessing eye images. Then, images are smoothed by an elliptical 2D-Gaussian filter which is constructed adaptive to the differences of iris characteristic scales, and code them by ordinal measure for regional grayscale changes between two sampling points.Finally, the Hamming distance is calculated between two iris codes and the recognition result is obtained according to that．Experimented on Bath and CASIA iris database, the proposed method has decreased the equal error rate to 0.05% and 0.69%. The results demonstrate that the proposed method has a high accuracy and fast identification speed.