In the era of mobile Internet, video advertisement is a common form among all ads. How to transfer consumers' attention to products well through video ads in the intense market environment is an issue that many companies focus on solving. By analyzing the components of the video, researchers believed that subtitles played an important role in conveying advertising content and images. However, some studies showed that subtitles might also obstruct the audiences' attention to the product. Therefore, the mechanism of the subtitles in the processing of video ads is still unclear and may be regulated by the characteristics of the products themselves. The product involvement is an important dimension. Product involvement is the degree to which an audience judges their relationship with an object based on their inherent needs, interests and values. The higher the degree of the relation, the higher the product involvement is. In this study, the method of eye movement technology is adopted to explore the influence of subtitles and product involvement on video ads processing through a 2-factor mixed design (Between subject variable is subtitle, the product involvement is within subject variable and the dependent variable is the eye movement data, including the first fixation duration, the number of fixation, the total fixation time, and the number of regression). At the beginning of the experiment, there were 30s of adaptive ads on the screen, followed by car advertisements. And then there were 10s blank on the screen. Next, beer ads are displayed. The order in the ads presented is balanced between the subjects. After the ad is over, the eye movement experiment is also ended. The eye tracker records the eye movement data of the subject automatically. The results suggest that subtitles and product involvement has significant impact on the processing of video ads: ads with subtitles generate more numbers of fixation and total fixation time than those without subtitles; high involvement product generate more numbers of fixation, total fixation time and numbers of regression than low involvement product, indicating that the processing of video advertisement with subtitles is superior to the video advertisement without subtitles and the processing of high involvement product is better than the low involvement product. Moreover, the interaction between subtitles and product involvement is significant: for high involvement product, the product with subtitles produces more numbers of fixation and total fixation time than the product without subtitles; there is no significant difference between low involvement product, indicating that the role of subtitles in video ads is affected by product involvement. The results show that subtitles affect the processing of video advertisements, but this effect is regulated by the degree of product involvement. Only the processing of high product involvement advertisements was affected by subtitles, also the results support the Elaboration Likelihood Model.