Background: Cocaine addiction is a global health issue with limited therapeutic options and a high relapse rate. Attentional bias towards substance-related cues may be an important factor for relapse. However, it has never been compared in former and current cocaine-dependent patients. Methods: Attentional bias towards cocaine-related words was assessed using an emotional Stroop task in cocaine-dependent patients (N = 40), long-term abstinent former cocaine-dependent patients (N = 24; mean abstinence: 2 years) and control subjects (N = 28). Participants had to name the colour of cocaine-related words, neutral words and colour names. We assessed response times using an automatic voice-onset detection method we developed and we measured attentional bias as the difference in response times between cocaine-related and neutral conditions. Results: There was an overall group effect on attentional bias towards cocaine, but no group effect on the colour Stroop effect. Two-by-two comparison showed a difference in attentional bias between cocaine-dependent patients and controls, whereas long-term abstinent former cocaine-dependent patients were not different from either. Although cocaine-dependent patients showed a significant attentional bias, consistent with the literature, neither long-term abstinent former cocaine-dependent patients nor controls showed a significant attentional bias towards cocaine-related words. We found no link between attentional bias size and either addiction severity or craving. Conclusions: Cocaine abstinence was associated with an absence of significant attentional bias towards cocaine-related words, which may be interpreted either as an absence of attentional bias predicting success in maintaining abstinence, or as attentional bias being able to disappear with long-term cocaine abstinence. Further research is needed to distinguish the role of attentional bias in maintaining abstinence.