A streaming media audio and video synchronous playing method based on B/S architecture comprises the following steps that A, at a sending end, the index number of a corresponding audio packet is introduced into the head of an RTP data packet of each video frame; b, the client receives the streaming media audio and video packets from the server side, caches the received audio and video packets intoa cache region, and sorts and frames the received audio and video packets; c, when the first index number corresponds to the second index number, synchronous processing is carried out; and D, the above steps are repeated to decode and output the video frame and the audio packet which are subsequently stored in the cache region until playing is finished. The method is simple and easy to operate and low in cost, the packet loss condition is processed without depending on a timestamp, a good smooth transition effect is achieved through the second-level cache, meanwhile, the processing and playing delay is reduced, and the lip sound inconsistency phenomenon caused by audio and video asynchronization in a real-time audio and video monitoring system is effectively eliminated.