Virtual worlds show promise for conducting meetings and conferences without the need for physical travel. Current experience suggests the major limitation to the more widespread adoption and acceptance of virtual conferences is the failure of existing environments to provide a sense of immersion and engagement, or of ‘being there’. These limitations are largely related to the appearance and control of avatars, and to the absence of means to convey non-verbal cues of facial expression and body language. This paper reports on a study involving the use of a mass-market motion sensor (Kinect™) and the mapping of participant action in the real world to avatar behaviour in the virtual world. This is coupled with full-motion video representation of participant’s faces on their avatars to resolve both identity and facial expression issues. The outcomes of a small-group trial meeting based on this technology show a very positive reaction from participants, and the potential for further exploration of these concepts.