In this paper, I propose an economic theory that addresses the epidemic character of opioid epidemics. I consider a community in which individuals are heterogenous with respect to the experience of chronic pain and susceptibility to addiction and live through two periods. In the first period they consider whether to treat pain with opioid pain relievers (OPRs). In the second period they consider whether to continue non-medical opioid use to mitigate cravings from addiction. Non-medical opioid use is subject to social disapproval, which depends negatively on the share of opioid addicts in the community. An opioid epidemic is conceptualized as the transition from an equilibrium at which opioid use is low and addiction is highly stigmatized to an equilibrium at which opioid use is prevalent and social disapproval is low. I show how such a transition is initiated by the wrong belief that OPRs are not very addictive. Under certain conditions there exists an opioid trap such that the community persists at the equilibrium of high opioid use after the wrong belief is corrected. Refinements of the basic model consider the recreational use of prescription OPRs and an interaction between income, pain, and addiction.