With the increasing bandwidth, videos have been gradually used as submissions for online peer assessment activities. However, their transient nature imposes a high cognitive load on students, particularly lowability students. Therefore, reviewers' ability is a key factor that may affect the reviewing process and performance in an online video peer assessment activity. This study examined how reviewers' ability affected the comments they provided and their reviewing behaviors and performance. Thirty-eight first-year undergraduate students participated in an online video peer assessment activity for 3 weeks. This study analyzed data collected from the teacher's and peer reviewers' ratings, comments provided by peer reviewers, and system logs. Several findings are significant. First, low-ability reviewers preferred to rate higher scores than high-ability reviewers did. Second, low-ability reviewers had higher review errors than high-ability reviewers. Third, high-ability reviewers provided more high-level comments, while low-ability reviewers provided more low-level comments. Finally, low- and high-ability reviewers showed different behavior patterns when reviewing peers' videos. In particular, low-ability reviewers invested more time and effort in understanding video content, while high-ability reviewers invested more time and effort in detecting and diagnosing problems. These findings are discussed, and several suggestions for improving the instructional and system design of online video peer assessment activities are provided.