Vertigo is a common disease, whereas the cause is very complex and wide-ranged. This indicates that high knowledge and skills are indispensable in diagnosis of vertigo. Regular doctor diagnoses vertigo in many cases, however they have little knowledge and skills about vertigo. In addition, the number of vertigo patient is increasing. From these reasons, the demand for supporting diagnosis of vertigo is growing. The purpose of this study is to develop an automated computer-aided diagnostic system of vertigo by video and image processing. One of the most important indicators in diagnosing vertigo is nystagmus, namely involuntary abnormal eye movement. For supporting regular doctor's diagnosis, the system must be easy to use and has high accuracy. Therefore, this paper focuses on analyzing nystagmus by videooculography( VOG) technique which is a video-based method of measuring eye movements using external infrared CCD camera. Previous study using VOG technique has mainly two problems, the noise from blink and the slow performance. This paper resolves these problems by proposal method for analyzing nystagmus. The proposal method can be divided into four stages: 1. detect blink, 2. estimate pupil position, 3. detect pupil position and radius, and 4. calculate rotation angle of torsional nystagmus. A total of 1000 images for each patient were used for evaluating the validity of proposed algorithm. As a result, noise from blink is completely removed in all patients. The proposed algorithm detects pupil in 100% accuracy in each patient, and detects the occurrence of torsional nystagmus. In conclusion, the results indicate that the proposed algorithm meets the requirements for supporting regular doctor.