With the recent development of image processing technology, the expression recognition from facial image has achieved human perception level . The need for expression recognition is not limited to analyze facial expression. By combining, the facial muscle information, it is possible to create analyze expression training system. There are two methods of expression training: seminar- and self-learning-based training. The self-learning methods is not effective, especially, it is difficult to determine the proper expression made by the users. The use of 3D morphing model many solve this issue because it can consider the facial muscle changes using fine muscles. This study proposes a new self-learning-base expression training system utilizing 3D morphing model. The proposed system started with capturing user’s facial image, and fitting a 3D morphing model to the image. Since the 3D morphing model contains parameters of six basic emotion, the facial image expression can be determining the fitting soon after the fitting process finished.