The development of computer technology provides a possible to carry out remote identification for biological quarantine specimens effectively, while at the present stage, most of the identification methods are lack of real-time interaction which makes the identification cost a lot of time. In this paper, we study the key technologies of web-based real-time remote expert identification, mainly in three parts: Online user detection using a self-designed finite state machine to solve the inherent defects of the web-based real-time controlling; real-time image transfer and display using Multi-threading which can improve the image transfer speed and long poling which can ensure receive images in real-time and Java script picture processing; web-based real-time audio and video interaction which is established by streaming media technology basing on flash media server. And then we implement the system. It does not require too much hardware support and has a strong web-based real-time controlling. It gets rid of geography constraints and can identify the biological specimens quickly.