Although numerous studies have demonstrated an association between PTSD and substance use disorders, little is known about the causal nature of this relationship. In this article, we put forth and test major causal hypotheses. Specific hypotheses to be tested include self-medication of PTSD symptoms, substance users' high risk of exposure to traumatic events, and drug users' increased susceptibility to PTSD following a traumatic exposure. We also examine the possibility of an indirect pathway linking drug use disorders and PTSD via a shared vulnerability. Evidence for these causal hypotheses is evaluated using Hill's criteria for causal inference: strength, consistency, specificity, temporality, gradient, plausibility, coherence, experimental evidence, and analogy. We present data analytic strategies that exploit information about the temporal order of PTSD and drug use disorders to shed light on their causal relationship. Finally, we present findings on the PTSD/drug use disorder association from an epidemiologic study of young adults.