A screen image observing device and method provides the sensations of virtual reality to the observer and reduces, to the extent possible, the discrepancy that is perceived by the observer between the movement of the observer's head and the displayed screen image. The device has a head tracking means, which detects the movement of the observer's head, and a screen image generating means, which generates image data based upon the result of the detection performed by the head tracking means. Image data is supplied to a display means.