Whether as medicine or for recreation, marijuana is used by relatively few Americans, yet, in recent years, a growing number of states have attempted to legalize the drug. To that end, this dissertation addresses three important puzzles related to marijuana legalization in the United States: 1) How did discourse around the issue of marijuana evolve? 2) How can we explain the rapid rise in the passage ballot initiatives dedicated to marijuana legalization? And 3) why are some places more supportive of legalization than others? The analysis presented in the dissertation focuses on the American states from 1990 to 2000; a period of increased political and discursive attention to marijuana. I develop a theory about how support for contentious issues evolves. First, I find that the marijuana issue has become increasingly characterized as a distributive rather than morality policy. Second, I find that electoral competition, liberal voting, and policy legacies contribute to quicker adoption of those legalization initiatives. Beyond these standard theoretical arguments, I demonstrate that legalization was adopted more rapidly in states where marijuana was reframed as a distributive policy -- places with higher positive discourse about marijuana and those with increasing discourse about marijuana's "revenue" benefits. Finally, this research also sheds light on the clustering of support for legalization. I argue that segregation of residents who may be more likely to oppose marijuana (e.g. parents) from others, creates a situation in which those groups come into less frequent contact, and thus develop lower stakes in the other's perception of marijuana. This spatial distribution of oppositional groups reduces the fears associated with the perceived negative consequences of exposure to marijuana, and increases local level support for legalization.