While legal scholars increasingly recognize that norms as well as law influence social behavior, the nature of these effects is not well understood. A key question concerns the content of norms. Specifically, do they reflect individual interest or do they enhance group welfare? In this paper I describe two general kinds of arguments that support these different views. I then develop predictions about the content of a particular type of norm—controller selection rules. These hypotheses are tested in an experimental setting in which individuals interact with each other over networked computers. The findings are consistent with group welfare explanations. That is, they suggest that norms constrain actions that create negative externalities, and that these norms are stronger in more cohesive groups. They also suggest, however, that such norms may not necessarily produce pro-social behavior.