On October 23, 2005, the Brazilian society faced a referendum with two options: "to not ban the legal market for firearms" and "to ban the legal market for firearms". Voters chose the first option. Along the months before the voting day, two interest groups were formed: one, called "No" - arguing for not banning the market - and the other group, called "Yes" - for the ban. This article has two main objectives: first, to analyze the arguments of each group using an analytical model and, two, to identify the determinants of the voter's choice using data at the local level. The main results indicate that the votes for No were negatively correlated to the homicide rate (measured in 2002) and positively correlated with the marginal change of the homicide rate (between 1995 and 2002. Thus, cities in which the violence increases more quickly, the citizens did not choose to ban the legal market for firearms.
