Reading plays an important role in our daily life, and eye tracking techniques have been extensively used to reveal important aspects of on line cognitive processing during reading. Over the past century, many scholars have conducted a great number of studies on the basic phenomenon and the underlying mechanism in the field of eye movements during reading. However, recently some new changes have taken place in terms of research topics, methods and techniques in this research field. How do researchers quickly capture and understand the main hotspots and trends of current research when they face so much knowledge? The mastery of these information will facilitate to carry out work efficiently, put forward new research questions, and further promote the development of eye movements in reading research. Many literature reviews are conducted through subjective analysis or meta-analysis, which to some extent, provide a good overview of the main content and research advances of a particular field. However, there are some limitations to these types of literature review, because they are not intuitive in the form of charts, and cannot include all the literature in this field. Therefore, bibliometric analysis is a good tool to carry out visualization on the literature of the eye movements in reading. In this study, the bibliometric analysis was used to examine the recent 20 years of research on eye movements in reading based on the original data of articles that were retrieved from “the Web of Science” database, and the index is “Science Citation Index Expanded” and “Social Sciences Citation Index”. The timespan was set from 2000 to 2021. The “eye movements” and “reading” were used as the topic words. What’s more, only articles and reviews were included, other document types (such as conference papers, abstracts, editorial materials, book chapters, letters, corrections and data papers) were excluded. Finally, a total of 3313 articles were included. CiteSpace 5.7.R2 was used as the tool for bibliometric analyses. The date of retrieval was March 6th, 2021. The preliminary results show that, firstly, publications of eye movements in reading have rapidly increased in the past two decades. Secondly, based on the Document Co-Citation Analysis and the timeline visualization of clusters, the E-Z Reader model and SWIFT model were the two influential computational models of eye movement control in reading. It is important to understand the critical controversies between the two models such as can readers extract semantic information from parafoveal words? do the linguistic properties of the upcoming parafoveal word influence processing of the currently fixated foveal word? Furthermore, the Linear Mixed-effects Models (LMMs) have been increasingly used in the field of eye movements in reading. Compared with traditional methods (e.g., ANOVA), the Linear Mixed-effects Models incorporate both fixed effects and multiple random effects, making data analyses more flexible and reliable. These advantages make LMMs become a standard method in the study of psychology of reading. Finally, keywords with the strongest occurrence burst and references with the strongest citation burst indicate that predictability will be a hot and flourishing topic in the near future. The question of how predictability effect occurs and affects at which stage of lexical processing (lexical access, a pre- or post-lexical stage) during reading requires further evidence. Predictability plays a vital role in constructing a comprehensive model of eye movement control in reading.