Three male research volunteers lived in a residential laboratory for 12 days in a study designed to investigate factors controlling patterns of marijuana smoking. All contact with the experimenters was through a networked computer system and subjects' behaviors were continuously recorded. During the first six hr of the day (0945-1545), subjects remained in their private rooms engaging in planned work activities, and during the remainder of the day (7 3/4 hr) they were allowed to socialize (1600-2345). Subjects were instructed that up to five active marijuana cigarettes (1.84% delta 9 w/w THC) could be smoked on designated days between 0945 and 2200. Cigarettes were available on request. Subjects requested all five cigarettes on 15 of 18 possible occasions (three subjects x six days of availability) with a mean latency to the first cigarette of 22 min. The pattern of self-administration was remarkably similar among subjects with all subjects smoking two cigarettes during the private work period and three cigarettes during the social access period. Subjects 1 and 2 smoked 90% of their social period cigarettes together in the social area, while Subject 3 smoked all of his cigarettes alone in his private room.