This invention provides a method and an apparatus for determining a rendering duration. An audio/video recording system is used for recording an audio bit stream and a video bit stream. The audio bit stream includes a plurality of audio frames. The video bit stream includes N video frames. According to this invention, the rendering durations of the first and the last video frames are determined based on the moments when the system starts to record the audio frames and the video frames. According to this invention, when the (i+1)th video frame through the (i+NL)th video frame are lost, the rendering duration of the ith video frame is determined based on NL. According to this invention, when the system receives a command of synchronization correction for the ith video frame, the rendering duration of the ith video frame is determined based on a correction value.