Social norms can influence behavior by establishing reward and punishment schemes that operate through community enforcement and in this way support outcomes that cannot be supported by personal enforcement alone. This paper analyzes the outcomes that can be supported through social norms in a society. I consider a society of infinitely long-lived and very patient agents that are randomly matched in pairs every period to play a given game and, unlike previous work that considers a society divided in two groups, I only restrict this matching procedure to be independent of history and time. I find that any mutually beneficial outcome can be supported by a self-enforcing social norm under both perfect information and a simple local information system. These Folk Theorem results explain not only how social norms can provide incentives to forestall opportunistic behavior and support cooperation in a community but also how they can support outcomes characterized by inequality. For example, self-enforcing social norms can support outcomes such as a "kingdom" (a society in which one member reaches the highest possible payoff) or a caste system (in which some groups of the population obtain higher payoffs than others).