Five groups of three healthy adult males (n = 15) participated in experiments designed to investigate the effects of social and behavioral context on human cooperative behavior and to determine whether marijuana effects on cooperative behavior were influenced by these factors. Using a time-based contingency, participation on lower probability activities during 3-, 4-, or 6.5-hr sessions was maintained by points that were exchanged for time to engage in higher probability activities. High- and low-probability activities were maintained by paying subjects different amounts of money for working on tasks. Sessions were conducted in socially isolated rooms, and each subject chose whether points earned while engaged in lower probability activities would be equally distributed to all subjects (cooperation) or kept by each subject individually (noncooperation). Subjects could monitor each others’ choices by observing illuminated panels in their rooms. The number of points available for choosing the noncooperative option was manipulated across sessions. In addition, social interaction between subjects varied across studies. Cooperative and noncooperative choice patterns were similar across sessions for all three subjects in each study, as were total session earnings. Under conditions in which subjects could not interact, or could interact only minimally, choices for cooperative and noncooperative point-earning options were directly related to the number of points associated with the options. However, when subjects were allowed 6.5 hr of social interaction per day, between successive sessions, only the cooperative option was chosen, regardless of point conditions. Marijuana altered activity performance, but had no effect on choice patterns, or on session earnings, under any condition.