This paper proposes a novel method for estimating temporally consistentdisparity maps. The proposed approach utilizes a coherencefunction that represents the temporal consistency between the disparitymaps of consecutive frames. The coherence function is computedaccording to a motion probability between consecutive framesbased on the assumption that a disparity varies smoothly over time.The temporal consistency between consecutive disparity maps is enforcedbased on the coherence function. In order to cope with scenechanges, the similarity of consecutive frames is also considered inthe coherence function. Experimental results show that the proposed method yields satisfactory disparity maps and synthesized views onseveral challenging stereo videos, including clips with fast objectand camera motions as well as scene change.