In standard drug self-administration settings, animals have no choice than drug use. As a result, serious doubt exists about the interpretation of drug use in experimental animals. Is it symptomatic of an underlying addiction state or merely an expectable response to lack of choice? This incertitude in turn casts a shadow over many behavioral and neurobiological changes that have been well documented in animals following extended drug self-administration. Do they reflect pathological dysfunctions or normal neurobiological adaptations? Here I address these questions by focusing on intravenous cocaine self-administration in the rat as a paradigm example. Overall, available evidence shows that when a valuable behavioral option, even a biologically or physiologically inessential one, is made available during access to cocaine self-administration, most rats readily abstain from cocaine use in favor of the alternative reward regardless of the amount of past cocaine use. Only a small minority of rats continue to self-administer the drug despite the opportunity of making a different choice. This pattern of results (i.e., abstinence in most rats; cocaine preference in few rats) maps well onto what is currently known about the epidemiology of human cocaine addiction. It is thus possible that the minority of cocaine-preferring rats would be homologous to the minority of human cocaine users with a diagnosis of addiction while the remaining majority of abstinent rats would be resilient to cocaine addiction. Choice could represent an objective method of selection of addicted animals for future research on the neurobiological dysfunctions that are hypothesized to underlie cocaine addiction. Other competing interpretations of the same pattern of results are also discussed at the end of this review. Copyright © 2010 Elsevier Ltd. All rights reserved.