Background Developing the means to identify smokers at high risk for relapse could advance relapse prevention therapy. We hypothesized that functional magnetic resonance imaging (fMRI) reactivity to smoking-related cues, measured before a quit attempt, could identify smokers with heightened relapse vulnerability. Methods Before quitting smoking, 21 nicotine-dependent women underwent fMRI during which smoking-related and neutral images were shown. These smokers also were tested for possible attentional biases to smoking-related words using a computerized emotional Stroop (ES) task previously found to predict relapse. Smokers then made a quit attempt and were grouped based on outcomes (abstinence vs. slip: smoking ≥ 1 cigarette after attaining abstinence). Prequit fMRI and ES measurements in these groups were compared. Results Slip subjects had heightened fMRI reactivity to smoking-related images in brain regions implicated in emotion, interoceptive awareness, and motor planning and execution. Insula and dorsal anterior cingulate cortex (dACC) reactivity induced by smoking images correlated with an attentional bias to smoking-related words. A discriminant analysis of ES and fMRI data predicted outcomes with 79% accuracy. Additionally, smokers who slipped had decreased fMRI functional connectivity between an insula-containing network and brain regions involved in cognitive control, including the dACC and dorsal lateral prefrontal cortex, possibly reflecting reduced top-down control of cue-induced emotions. Conclusions These findings suggest that the insula and dACC are important substrates of smoking relapse vulnerability. The data also suggest that relapse-vulnerable smokers can be identified before quit attempts, which could enable personalized treatment, improve tobacco-dependence treatment outcomes, and reduce smoking-related morbidity and mortality.