Craving is a commonly used term to describe an intense desire for a substance or behaviour; however, its underlying neurobiology is not fully characterized. We have successfully used a cue exposure paradigm with functional neuro-imaging ((H2O)-O-15 PET; PET, positron emission tomography) in abstinent opiate addicts. This study showed that salient cue exposure results in activation in the left anterior cingulate/mediofrontal cortex and elicited craving correlated with activity in the left orbitofrontal cortex. We therefore aimed to replicate this study in alcohol dependence to see if a similar pattern of neural activation occurred. We recruited six abstinent alcohol-dependent and six non-dependent subjects who each underwent a 12-run PET scan using (H2O)-O-15 to measure changes in regional blood flow during exposure to an alcoholic drink or its visually matched non-alcoholic drink. Physiological data and subjective ratings were also recorded. Statistical parametric mapping (SPM99) was used to analyse the PET images. Compared with control subjects, abstinent alcohol-dependent subjects rated their alcohol craving higher at baseline and throughout the study, but there was no significant change in the scores in response to the cues in either group. SPM analysis across all subjects showed significant activation in the occipital cortex in response to the alcohol cue as compared with the neutral one. Analysis of the same regions that were activated in the opiate study, revealed significant increases in signal activation in the left medial prefrontal area, but only in abstinent alcohol-dependent subjects. In conclusion, in abstinent alcohol dependence we suggest that a simple cue exposure paradigm is not sufficiently powerful in functional imaging studies to determine the underlying neurobiology of subjective craving. Comparisons with the finding in opiate dependence suggest a shared region, the anterior cingulate/left medial prefrontal cortex is involved in the cue response in dependent subjects but not controls.