The invention discloses a three-dimensional video encoding method. The method has the advantages that: the encoding quality of an image is changed by changing a quantization parameter, thus changing the code rate of the image, so as to adapt to the changing of the network transmission bandwidth; under the condition of high bandwidth, a left-viewpoint basic-layer code stream, a left-viewpoint enhancement-layer code stream, a right-viewpoint basic-layer code stream and a right-viewpoint enhancement-layer code stream are transmitted at the same time, thus acquiring high-quality videos; under the condition of low bandwidth, however, only the left-viewpoint basic-layer code stream and the right-viewpoint basic-layer code stream are transmitted, a client can also normally decode and acquire low-quality videos, thus leading good network adaptability of the method; as an automatic three-dimensional display device can synthesize the video which is corresponding to the present viewing angle by the existing viewpoint video when the viewing angle is changed, namely the virtual viewpoint synthesis. The method leads an automatic three-dimensional display terminal of users to be capable of quickly realizing the virtual synthesis of multi-viewpoint images with low complexity by providing the parallax field code stream for the client, thus satisfying the requirement of real-time automatic three-dimensional displaying.