Computer games are always played in particular places by particular people, but this fact is often -glossed over- in order to build general accounts of what computer gaming is and what knowledge the player might acquire through playing. This study explicates some features of the immanent pedagogies of co-located computer gaming, utilizing an approach in which interaction analysis is used to explore the details of gaming practices in an internet café. Recordings of on- and off-screen data was obtained during fieldwork in such a location, capturing the various semiotic resources that participants use to interact, including verbal and corporeal actions as well as on-screen actions such as cursor pointing and avatar movement. The study focuses specifically on instructional sequences in cooperative game playing, exemplifying these practices through the display of two gaming sessions, one of equally competent players and one where one of the players is a complete novice to this game. Differences and similarities between these two constellations are highlighted. In sessions where players have a marked difference in skill, instructions often take the form of direct imperatives, in which a more competent player will order the one being instructed to do some specific action. Rather than trying to get the less competent player to reflect upon actions, instructions are geared to demonstrate concrete sequences of on-screen actions. In the session with equally skilled players, instructions are more subtle, as knowledge of the game-s mechanics and game tactics is assumed. Common to both these constellations of players is the way in which instructions are focused on getting players to act in ways directly relevant to the current game-state. Instructions also serve to demonstrate the professional vision of gaming, and practices of concerted highlighting relevant aspects of the game-s interface are commonplace.