Chinese characters are comprised of a variety of strokes with some simple features such as dots, lines, and curves. Are all strokes within a character equally important in the character identification? There are some studies to explore this field. The preliminary results showed that beginning strokes play a more central role in character identification than ending strokes. And it was clearly the case that certain strokes are more central to successful character identification than the others. To explain such effect, a hypothesis has been put forward in the present study, “The order of writing affects the identification.” The present study was based on the study of Yan et al. (2012). But a more sensitive and accurate indicator was chose to investigate the role of stroke encoding in character identification. Instead of the number of the strokes, the present experiment tried to use the “pixel”, the basic unit made up a stroke, to examine the effect. This experiment is a 2(Removal Type: Beginning, Ending) ×3(Percentage of Pixels Removed: 15%, 30% and 50%) within-subject design. In addition to these six experimental conditions, a control condition was included in which no strokes were deleted from the characters. Thirty-five undergraduate students participated in the experiment. Participants’ eye movements were recorded with a SR Research EyeLink 2000 eye tracker. Each participant was tested individually. Global analyses were conducted based on measures of reading behavior across the whole sentence. Repeated measures ANOVAs and pairwise comparisons were carried out. All the measures showed that Chinese characters with 15% of pixels removed were as easy to read as Chinese characters without any pixels removed. However, when 30%, or more of a character’s pixels were removed, there was a robust effect in relation to the percentage of pixels removed: readers made more and longer fixations, more regressions and forward saccades, and longer total reading time when reading text with increased percentages of pixels removed. Thus, it appears that text with more pixels removed produced more disruption to Chinese reading. Also, there was a robust effect on pixel removal type. Characters with beginning pixels removed were more difficult than that with ending pixels removed. The results suggested that not all strokes within a character are of equal importance in the character identification processing. That is, a stroke is laid down during Chinese character production, to some extent at least, determines how important that stroke is in relation to character identification in reading. Strokes that are laid down early during written character production appear to be more important than strokes laid down late. And to the most important, these results supported our hypothesis that the order of writing affects the identification both in English and Chinese.