Embodiments of the present invention disclose a method, terminal and system for caption transmission for telepresence, relating to the field of telepresence technologies and solving the problem that caption display on each display terminal cannot be coordinated in the prior art. Some embodiments include acquiring entire caption content and attribute information corresponding to the caption content; determining, according to the entire caption content and the attribute information corresponding to the caption content, caption content to be displayed on each display terminal and a display timing; and transmitting the determined caption content to each corresponding display terminal according to the display timing so that each display terminal superposes the determined caption content to an image.