BACKGROUND AND OBJECTIVES:In order to analyze and understand the oculomotor system completely, it is essential to measure horizontal, vertical, and torsional eye movements. We developed a new system of 3-dimensional videooculography that can accurately estimate the pupil center and measure torsional eye movement in real time while minimizing the errors caused by upper eyelid droop, eyelashes, corneal reflection, and eye blinking. MATERIALS AND METHOD:For accurate estimation of the center and diameter of the pupil, the edges of the pupil were searched by using a sector-form window and circle fitting was performed using the least square regression. Torsional eye movement was measured using polar cross-correlation technique with modification to lessen the errors produced by the change in pupil size. Calibrations were performed and to verify the proposed system, the linearity between the measured and actual eye movements was measured. Also, the results of the proposed system were compared with those of another commercialized system. RESULTS: The linearity between the measured and actual eye movements showed errors of less than 1%. The spatial resolutions of the system for horizontal, vertical, and torsional eye movements were 0.3, 0.4, and 0.2, respectively whereas the temporal resolution was 30 frames/sec. In comparison to the commercialized system, our system showed less artifact by eye blinking when measuring vertical eye movement. CONCLUSION:We proposed an algorithm and a system for measuring horizontal, vertical, and torsional eye movements which minimized the influence of partial eyelid closure, eyelashes, corneal reflections, and change in pupil size.