The co-occurrence of substance use disorders (SUDs) and anxiety disorders has been now well established. This association is frequent and can be explained by three models: the shared vulnerability factors model, the self-medication model, and the substance-induced model. General population epidemiological studies provide strong evidence of the frequency of the association for the most used substances: tobacco, alcohol, cannabis, and to a lesser extent sedatives, opiates, and cocaine. For substances that are less commonly used in the general population, the frequency of the co-occurrence can more precisely be studied in clinical samples. We provide the most recent literature results on the association of SUDs and anxiety, and evidence for one explicative model or the other when available. For substances with sedative properties (alcohol, benzodiazepines, cannabis, opioids), both evidence for a self-medication and for a toxic effect exist. For substances with psychostimulant properties (tobacco, cocaine, and amphetamines), the literature favors the toxic hypothesis to explain the association with anxiety disorders. We give practical steps for the recognition of these dual diagnoses and present therapeutic issues, although the strategies are rarely evidence based.