In the process of reading, the location and duration of eye fixations varied accroding the language properties of the text. Using the eye tracker nowdays, it precisely records the location and fixation duration of eye movements. That gives an opportunity to understand the cognitive operations of reading. Furthermore, the eye movement contingent display technique is capable to change the display with both the speed and precision. It changes the text on the computer screen at specific location and time, according the location where the eyes are fixating. This advanced technique provides the possibility to explore the nature of dynamic processing of reading. Experiment One uses the moving window paradigm to measure the perceptual span of Chinese reading. It showed that the range of perceived characters in reading includes four characters to the right of fixations. Experiment Two also uses the moving window paradigm but the time of showing "window" is varied. It showed that the characters in peripheral were perceived within 90 msec in the beginning of fixations. The time to mask the peripheral text did not affect the fixation duration of display change, but the time of next fixation. Experiment Three uses the boundary paradigm to show the type of character information can be retrieved in peripheral and integrated with the information on the next fixation. It showed the phonological code can be retrieved in peripheral, but the orthographic code is dominated when integrating information across fixations. Experiment Four uses the fast priming paradigm to show the type of information can be retrieved on fixations. It showed the same pattern for both the third and forth experiment. The phonological effect can be obtained under certain circumstances, when the availability or validity of orthographic information is reduced. The dynamic processing of eye movements in reading and a general model for cognitive operation are discussed.