Identifying addicts with higher risk of relapse would provide the opportunity to implement individualized interventions and increase cessation success rates. Unfortunately, the ability to predict the long-term success of drug cessation treatments continues to elude researchers. We tested whether brain responses to emotional and cigarette-related pictures were predictive of the ability to abstain from smoking. Smokers interested in quitting (n = 180) participated in a smoking cessation clinical trial. Before the initiation of any treatment, we recorded event-related potentials (ERPs) evoked by emotional (both pleasant and unpleasant), neutral, and cigarette-related images. Cluster analysis was used to assign smokers to two groups based on the amplitude of the late positive potential (LPP) to the experimental stimuli. While both groups showed enhanced responses to cigarette-related cues, one group (n = 81) also showed blunted brain responses to intrinsically pleasant stimuli. Smokers in the latter group were significantly less likely to be abstinent at 10, 12 and 24 weeks after their quit date. In conclusion, using ERPs, a direct measure of brain activity, we found that smokers with blunted brain responses to intrinsically pleasant stimuli had lower rates of long-term smoking abstinence. This response offers a new biomarker for identifying smokers at higher risk of relapse and for testing the efficacy of new interventions aimed at normalizing brain reward systems' responses to intrinsically pleasant stimuli.