Method and system for discriminating harmful link in instant message

The invention is used for automatically identifying harmful links in real time message. The method thereof comprises: a) building links database at server side, and then setting the identifying information respectively corresponding to the harmful links or harmless links; b) when user clicks the links in the real-time message, the webpage browser directly transmits the links to the server side; the server side will decides if the links are harmful; if yes, then prompting user that the links are harmful using webpage browser; otherwise, the webpage browser executes the link.