In current online teaching system,the users havecertain blindness on choosing teaching videos,according to this characteristic,online teaching video recommendation method based on collaborative filtering is pointed out,which can recommend the teaching videos in which the users are interested to other users.Firstly,the video-watching records are saved indatabasse,and neighboring user are searched based on historic viewing records of each user and the interest difference between each user in basic information,then teaching video recommendation is conducted acording to video-watching records of these neighboring users based on collaborative filtering method.This method improves the Sparse and Cold Start commonly existed in traditional collaborative filtering recommendation method and makes th erecommendation more accurate.In addition, wherher the users watch the recommended videos can make implicit evaluation on the system to revise the parameters of the system and to improve the acccuracy of the recommendation.