The present invention relates to a device and a method for coding a 3D video, a decoding method, according to the present invention, comprising the steps of: receiving a single depth mode (SDM) flag information indicating whether the SDM is applied to a current block; and if the SDM is applied to the current block, generating a prediction sample on the basis of a first candidate and a second candidate derived by using spatial neighboring reference samples of the current block, wherein the number of the spatial neighboring reference samples is two, and among the neighboring reference samples, the first neighboring reference sample is positioned on the left side of the current block, and the second neighboring reference sample is positioned on the upper side of the current block. According to the present invention, the current block may be predicted on the basis of one representative value derived by using the predetermined neighboring reference sample, and the complexity of an operation may be decreased, and coding efficiency may be increased.