3D stereoscopic technology causes visual discomfort and several studies have attempted to assess the degree of discomfort using EEGs. However, there is a lack of scientific basis for the selection of electrodes during the assessment. The study aims to find representative electrodes for assessing stereoscopic visual comfort via stereo vision brain mechanism analysis. The study reports measurements of EEG activities of 15 electrodes within stereo vision associated brain regions, and the temporal components and frequency bands of the EEG data were analyzed using event-related potentials and power spectrum analysis to find representative electrodes that differentiate between visual comfort and discomfort states. The results showed that the peak amplitude and band power of the F7, Cz, P4, and O2 electrodes were significantly different (p < 0.05) between the visual comfort and discomfort states, and these four electrodes are called representative electrodes. Meanwhile, the classification accuracy of all electrodes over representative electrodes was only improved by 2-4 % (p > 0.05) for both machine learning and deep learning models.