At university, students are often confronted with a synthesis writing task. Through writing synthesis texts, in which information from sources is integrated, students direct their learning. Collaborative writing often leads to better synthesis texts than individual writing. Although collaborative writing can improve students' learning outcomes, the process leading to these outcomes deserves more attention. This study aims to elicit how university students collaboratively write a synthesis text in terms of their overall approach, their specific interactions, their collaboration skills and their group atmosphere. Two groups of three master's students in educational studies wrote a synthesis text in Google Docs within three hours, based on provided sources. They communicated via Zoom and their interactions were recorded. Based on these recordings, the overall approach was determined. In the two transcripts, units of meaning were coded using an adapted version of the coding scheme for productive interaction by Damşa [1]. Collaboration skills were investigated via questionnaires. Group atmosphere was determined by three independent coders based on the Fiedler Group Atmosphere Scale [2]. Results show that the two groups differed in terms of overall approach, interactions, collaboration skills and group atmosphere. Based on these two groups, it can be concluded that collaboratively writing synthesis texts is not selfevident. Further research should focus on how to support collaborative writing processes.