Online discussions have been widely utilized as an educational activity, and much research has been conducted on the process and behaviors involved in asynchronous discussions. However, research on behavioral patterns in learners' synchronous discussions, including the process of social knowledge construction and project coordination is limited. Through the examination of the behavioral patterns and differences between students with high- and low-quality discussions, it may be possible to understand the limitations of knowledge construction in synchronous discussions. Furthermore, these findings may help teachers design and guide synchronous discussions activities. This study is an empirical case study in which college students conducted synchronous discussions based on topics specified by the teacher. The students used a text-based instant messaging (IM) tool in a period of 98 days. Two analytical methods were employed. The coding of the discussion messages was followed by a quantitative content analysis and a lag sequential analysis of behaviors. The social knowledge construction, project coordination, and social interactions in the group discussion were explored. Furthermore, the differences between the behavioral patterns of the high- and low-quality discussion groups were also examined. The findings indicate that although more than half of the discussion messages were off-topic, there were also some knowledge construction behavioral sequences. Furthermore, there were several limitations on the diversity and depth of the knowledge construction in the students' discussions. The high-quality discussion teams performed better than the low-quality discussion teams in terms of participation, diversity in knowledge construction, and coordination. However, they also had more off-topic discussions. In this paper, we discuss these behavioral patterns and provide specific suggestions for teachers regarding how to implement synchronous discussions that are targeted to students' knowledge construction processes. (Contains 5 tables and 2 figures.)