A video frame decoding method and apparatus in order to implement efficient and fast video decoding. The method includes acquiring a target time point of a video, obtaining a frame number of a corresponding random access frame according to the target time point, obtaining a frame number of a frame that can be independently decoded, where the dependency set includes at least one dependency, and the dependency represents a correspondence between a frame number of a random access frame and a frame number of a frame that can be independently decoded, and obtaining, according to the frame number of the frame that can be independently decoded, video data of the frame that can be independently decoded, and decoding the random access frame.