In the Conferencing and Collaboration (COCO) group within SMLI, we have developed a prototype desktop conferencing system that provides audio, video, and shared drawing connections among workstation desktops. In order to come to grips with some of the many issues involved in introducing multimedia and collaborative technology to the desktop, we have been involved in several studies of collaborative activity. We did a survey of the users of the commercial video conferencing rooms used within Sun to assess users' perceptions of what they needed to support remote collaboration. We analyzed videotapes of an actual working team that was split between East and West Coast sites to understand their actual collaborative activity. We are currently in the process of using our desktop conferencing prototype to study a distributed working team under 3 conditions: using conventional collaboration tools, using desktop conferencing, and using desktop conferencing without video. In this talk, I'll be presenting the results of these studies (although at the time of talk, I'll only be able to talk about the first 2 conditions of the last study). I'll discuss not only the findings of the studies, but some of the methodological issues of using different approaches to collect and analyze data on work practice.