The present study extends the tradition of research on individual vs. group risk taking to the case in which the decisions have consequences for persons not implicated in the act of decision-making. The decision task consisted of gambles with real monetary payoffs. Ss--157 French female university students--were randomly assigned to one of four conditions: (1) a control group in which decisions were made for oneself on two occasions; (2) decisions made individually, first for oneself and then for another person; (3) individual decisions made for oneself followed by group discussion to consensus, with decisional consequences impinging only upon the decision-making group; (4) individual decisions made for another person followed by group discussion to a consensus affecting another group. Condition 2 yielded conservative shifts; Condition 3 produced risky shifts. Condition 4 shifts were in the risky direction, but of lesser strength than those obtained in Condition 3. Interpretation of the results considered the possible applicability of social responsibility and reciprocity norms to the phenomena under investigation.