The basic concepts underlying compulsive, impulsive and addictive behaviours overlap, which may help explain why laymen use these expressions interchangeably. Although there has been a large research effort to better characterize and disentangle these behaviours, clinicians and scientists are still unable to clearly differentiate them. Accordingly, obsessive-compulsive disorder (OCD), impulse control disorders (ICD) and substance-related disorders (SUD) overlap on different levels, including phenomenology, co-morbidity, neurocircuitry, neurocognition, neurochemistry and family history. In this review we summarize these issues with particular emphasis on the role of the opioid system in the pathophysiology and treatment of OCD, ICD and SUD. We postulate that with progression and chronicity of OCD, the proportion of the OCD-related behaviours (e.g. checking, washing, ordering and hoarding, among others) that are driven by impulsive 'rash' processes increase as involvement of more ventral striatal circuits becomes prominent. In contrast, as SUD and ICD progress, the proportion of the SUD- and ICD-related behaviours that are driven by compulsive 'habitual' processes increase as involvement of more dorsal striatal circuits become prominent. We are not arguing that, with time, ICD becomes OCD or vice versa. Instead, we are proposing that these disorders may acquire qualities of the other with time. In other words, while patients with ICD/SUD may develop 'compulsive impulsions', patients with OCD may exhibit 'impulsive compulsions'. There are many potential implications of our model. Theoretically, OCD patients exhibiting impulsive or addictive features could be managed with drugs that address the quality of the underlying drives and the involvement of neural systems. For example, agents for the reduction or prevention of relapse of addiction (e.g. heavy drinking), which modulate the cortico-mesolimbic dopamine system through the opioid (e.g. buprenorphine and naltrexone), glutamate (e.g. topiramate), serotonin (e.g. ondansetron) or γ-aminobutyric acid (e.g. baclofen and topiramate) systems, may prove to show some benefit in certain forms of OCD. Based on the available evidence, we suggest that the treatment of patients with these disorders must account for alterations in the underlying motivations and neurobiology of the condition. We provide an initial guide to the specific treatments that future clinical trials might consider in patients with OCD. For example, it might be wise to test naltrexone in patients with co-morbid SUD and ICD, topiramate in patients with co-morbid ICD and eating disorders, and baclofen in patients with co-morbid Tourette's syndrome. These trials could also include scales aimed at assessing underlying impulsivity (e.g. Barratt Impulsiveness Scale) to check whether this construct might predict response to drugs acting on the reward system.