Multiple transitions between ‘fiction’ and ‘reality’ have been discussed as an interesting feature of players’ practices. However, the detailed ways in which players situatedly switch in and out of the game have not yet been described. Based on a video-recorded corpus of virtual football games, this paper describes the systematic organisation of players turns-at-talk when they are deeply involved in the game, and shows how turns are transformed as they disengage from the game. These two ways of formatting turns are characterised by two forms of temporality: the first is embedded in the game and is shaped by the mobility and speed of the game dynamics; the second is shaped by the interactional space created between the co-present players. The paper shows how these two regimes are organised in an embodied way: in the former, players are visually focused and bodily oriented towards what happens on their TV monitor; in the latter, players are mutually oriented, exchange glances and gesticulate one towards the other.