Deep brain stimulation of the posterior hypothalamus is a therapeutic approach to the treatment of refractory chronic cluster headache, but the precise anatomical location of the electrode contacts has not been clearly assessed. Our aim was to study the location of the contacts used for chronic stimulation, projecting each contact centre on anatomic atlases. Electrodes were implanted in a series of 10 patients (prospective controlled trial) in the so-called 'posteroinferior hypothalamus' according to previously described coordinates, i.e. 2 mm lateral, 3 mm posterior and 5 mm below the mid-commissural point. The coordinates of the centre of each stimulating contact were measured on postoperative computed tomography or magnetic resonance imaging scans, taking into account the artefact of the electrode. Each contact centre (n=10; left and right hemispheres pooled) was displayed on the Schaltenbrand atlas and a stereotactic three dimensional magnetic resonance imaging atlas (4.7 tesla) of the diencephalon-mesencephalic junction for accurate anatomical location. Of the 10 patients with 1-year follow-up, 5 responded to deep brain stimulation (weekly frequency of attacks decrease >50%). In responders, the mean (standard deviation) coordinates of the contacts were 2.98 (1.16) mm lateral, 3.53 (1.97) mm posterior and 3.31 (1.97) mm below the mid-commissural point. All the effective contacts were located posterior to the hypothalamus. In responders, structures located <2 mm from the centres of effective contacts were: the mesencephalic grey substance (5/5), the red nucleus (4/5), the fascicle retroflexus (4/5), the fascicle longitudinal dorsal (3/5), the nucleus of ansa lenticularis (3/5), the fascicle longitudinal medial (1/5) and the thalamus superficialis medial (1/5). The contact coordinates (Wilcoxon test) and the structures (Fisher's exact test) were not significantly different between responders and non-responders. These findings suggest that failure of deep brain stimulation treatment in cluster headache may be due to factors unrelated to electrode misplacement. They also suggest that the therapeutic effect is probably not related to direct hypothalamic stimulation. Deep brain stimulation might modulate either a local cluster headache generator, located in the hypothalamus or in the mesencephalic grey substance, or non-specific anti-nocioceptive systems.