In Chinese character recognition, the relationship between global effect and local effect is a hotly debated issue for researchers. However, there has been inconsistent hypotheses and results among previous studies. The current study is designed to investigate the priority of global effect and local effect in the lexical decision task by using both reaction time and fixation time as indicators. This study used a 4 (orthography: correct, false radical-location, false radical-component, and scramble-stroke) × 2 (font size: 1.8°, 18°) repeated-measure design. We selected 128 Chinese target characters (half left-right structure and half upper-lower structure) to create the experimental stimuli. The false radical-location stimuli were made by reversing the two components of the correct characters; The false radical-component stimuli were made by adding or reducing a stroke to a specific radical in the correct characters; The scramble-stroke stimuli were made by a random rearrangement of the 3x3 matrix division for the correct characters. Each stimulus had a large size condition and a small size condition. The experimental conditions were counterbalance across participants. We also added correct characters as fillers in order to balance the number of right and wrong items. As a result, each participant was presented 128 conditional characters and 64 fillers. An EyeLink ELCL desktop system was used to monitor participants’ eye movements. When a trial started, a central fixation point was presented in the central of screen for 500 ms. Then a stimulus was presented in the center of the screen. The participant was required to press a button on a joystick to signal whether the stimulus was a real and correct character or not. The left and right buttons were assigned as yes- and no-response respectively in half participants, and vise visa in the other half participants. The presentation of stimuli would be stopped immediately by button pressing within the 2000 ms timeout. The results show that the false radical-location condition was significantly shorter than the false radical-component condition in reaction time, regardless of font size (ps < .001). In first-fixation duration, there was no significant difference between the false radical-location and the false radical-component conditions for small font size (p = .183), and the false radical-component condition produced longer fixation duration than the false radical-location condition for large font size (p = .036). In the total time, the false radical-component condition produced shorter fixation duration than the false radical-location condition for both font sizes (ps < .001). We conclude that while the result of reaction time supports the dominance of global effect in lexical decision, the result of fixation duration indicates that there is a time course of global effect and local effects in lexical decision. The eye movement technique may be a useful way to investigate the Chinese character processing in future research.