Based on some priori knowledge such as the feature information of prostate magnetic resonance images (MRIs) and its special zone different diseases often occur, aiming at the segmentation of prostate outer and inner contours, a twostep prostate MRI segmentation method based on edge distance regularized level set evolution (DRLSE) is proposed. On the basis of constructing a unified level set energy function, the first step is to realize the outer prostate contour segmentation based on the T1 (longitudinal relaxation time) image of MR. The second step is to realize the inner prostate contour segmentation based on the T2 (transverse relaxation time) image of MR under the condition of outer contour constraint; and then the effective whole segmentation of the prostate outer and inner contours is achieved. The humanmachine interactive interface of the prostate segmentation was designed, and the image segmentation experiment study on ten prostate cases (including 30 images of normal, hyperplasia and cancerous prostates) was conducted. The Dice similarity coefficient (DSC) is used to evaluate and analyze the segmentation results, and the DSC value reaches to above 90%. The experiment results show that the proposed twostep prostate MRI segmentation method based on edge DRLSE can effectively realize the whole segmentation of prostate outer and inner contours, and the results are very close to the ideal ones obtained by clinic experts manually, which has good reference value for the clinical diagnosis and treatment of prostate diseases.