Abstract: In order to help the handicapped and the elderly use computer for information interacting, a low-cost and non-contact eye gaze tracking system is designed. This system is composed of two-loop illuminators and a CCD camera. The way of getting pupil position is to do subtraction of the two images, "bright pupil" and "dark pupil", which are gained by illuminating user's eyes alternate. The pupil center is localized by ellipse fitting, and the Purkinje image can be obtained in the "dark pupil" image, then the local gazing direction can be determined. Based on the characteristic of human gazing movement, the tracking strategy of pupil that is designed by neural network and Kalman filtering is discussed. At last, an eye gaze mapping model and its calibration are given.