This thesis investigates the phenomenon of social interaction in shared virtual environments (SVEs), supported by virtual reality (VR) systems over time. SVEs are computer generated 3D graphical spaces where geographically distributed people can meet and interact with each other in a graphical space. Although there have been a number of studies about social interaction in SVEs, there has been a lack of research looking into changes over time, which this thesis does. In order to gain more knowledge about social interaction over the longer term, this thesis compares and contrasts four different types of VR systems that supported various SVEs. Two of the systems were internet based SVEs on desktop computers where many users could interact at the same time. One of the SVEs had voice based communication. The other SVE had text based communication. The other two were based in laboratory settings. One setting was networked immersive projection technologies (IPT) in which two participants performed a variety of tasks together. The other was one IPT connected to a desktop VR and participants changed systems half way through the trial in which they collaboratively solved a task together. In both settings voice based communication were used. Observations and other methods of analysis were carried out, focusing on differences and similarities in peoples behaviors in the process of social interaction over time in SVEs. The six papers contained in this thesis explore social interaction over time in shared virtual environments. This thesis argues that technology becomes not only a tool for social interaction; it also becomes a key aspect in social interaction. While the technology filters out some of the social cues we are familiar with from face to face situations, it also ‘filters in’ new cues that become important for how people can connect to each other in the shared virtual environment. Over time, these social cues, that people creates among themselves while using the technology, become essential for people learn about; otherwise they find it difficult to relate to each other and do things together in the shared virtual environments. The more difficulties people have in figuring out how to use the technology while interacting with others, the less they will accept the technology as an appropriate tool for connecting people and doing things together. The reason for this is that social and technical issues can only be separated analytically in shared virtual environments; in practice, as this thesis shows, they are highly intertwined. This thesis puts forward a dynamic model identifying the importance of looking more explicitly at individuals, technology, task and time while studying social interaction in SVEs. In this way, the thesis combines a number of insights both from previous social science theories of social interaction and practices - together with observations from the studies this thesis builds on. The thesis puts forward a concept that includes these insights - connected practice, defined as the dynamics of social interaction in technical systems. This concept can guide future studies to incorporate both technical and social aspects over time since it was shown to be the key to understanding the phenomenon of this thesis. It is finally suggested in the thesis that the concept connected practice can be utilized in other technical systems apart from SVEs in future research of social interaction in technical systems.