The invention discloses a video cloud on-demand cache scheduling method supporting a multi-code-rate version. Considering the correlation of versions with the same video content but different code rates, a cache group share scheduling algorithm is adopted to reduce the cache resource expenditure. According to the method, for the same video content, video files of multiple code rate versions are stored in a streaming media server to adapt different requirements of a user; when the user requests the video file of a certain code rate version, direct service is performed if the file of the code rate version has been cached in the cache server; whether a file with an adjacent code rate version is cached is looked up in the cache server if the file of the code rate version is not cached; and if the file of the code rate version is cached, the request is firstly added into the cache group to reduce the wait time length of the user.