In remote tabletop collaboration multiple users interact with the system and with each other. Thus, two levels of interaction human-computer interaction and human-human interaction exist in parallel. In order to improve remote tabletop systems for multiple users both levels have to be taken into account. This requires an in-depth analysis achieved by qualitative methods. This paper illustrates how a combination of Activity Theory and Grounded Theory can help researchers and designers to improve and develop better collaborative interfaces. Findings reported here are based on three video recordings that have been collected during a quasi-experiment.