An interaction method based on multimedia programs, a terminal device, a server and a system are disclosed, which belong to the field of network technologies. The method includes: acquiring at least one image clip information of a multimedia program in a process of playing back the multimedia program; sending the image clip information, the image clip information being used for matching corresponding multimedia program information; receiving interaction information corresponding to the multimedia program information; and outputting an interactive interface, to interact with the multimedia program based on an interaction operation on the interactive interface. The present disclosure is easy to operate, and improves interaction efficiency.