A key reason for using asynchronous computer conferencing in instruction is its potential for supporting collaborative learning. However, few studies have examined collaboration in computer conferencing. This study examined collaboration in six peer review groups within an asynchronous computer conferencing. Eighteen tertiary students participated in the study. Content analyses of discussion protocols were performed in terms of participation, interaction, and social presence. The results indicate that collaboration does not occur automatically in asynchronous computer conference. Collaboration requires participation because no collaboration occurred in the two groups with low student participation; however, participation does not lead to collaboration, evidenced by student postings receiving no peer responses. Collaboration requires interaction but does not end with interaction, substantiated by different levels of collaboration across different interactional patterns. Social presence helps to realise collaboration through establishing a warm and collegial learning community to encourage participate and interaction, exemplified by the contrast of the group with the highest level of social presence and the group with the lowest level of social presence. A model of understanding and assessing collaboration in online learning is recommended, consisting of participation, interaction and social presence.