The availability of commodity depth sensors such as Kinect have enabled development of methods which can densely reconstruct arbitrary scenes. While the results of these methods are accurate and visually appealing, they are very often incomplete. This is either due the fact that only part of the space was visible during the data capture process or due to the surfaces being occluded by other objects in the scene. In this paper, we address the problem of completing and refining such reconstructions. We propose a method for scene completion that can infer the layout of the complete room and the full extent of partially occluded objects. We propose a new probabilistic model, Contour Completion Random Fields, that allows us to complete the boundaries of occluded surfaces. We evaluate our method on synthetic and real world reconstructions of 3D scenes and show that it and qualitatively outperforms standard methods. We created a large dataset of partial and complete reconstructions which we will make available to the community as a benchmark for the scene completion task. Finally, we demonstrate the practical utility of our algorithm via an augmented-reality application where objects interact with the completed reconstructions inferred by our method.