Speckle reduction of retinal optical coherence tomography (OCT) images helps the diagnosis of ocular diseases. In this Letter, we present a speckle reduction method based on shrinkage in the contourlet domain for retinal OCT images. The algorithm overcomes the disadvantages of the wavelet shrinkage method, which lacks directionality and anisotropy. The trade-off between speckle reduction and edge preservation is controlled by a single adjustable parameter, which determines the threshold in the contourlet domain. Results show substantial reduction of speckle noise and enhanced visualization of layer structures as demonstrated in the image of the central fovea region of the human retina. It is expected to be utilized in a wide range of biomedical imaging applications.