For almost one hundred years the neurocircuitry that underpins the cardiovascular and respiratory responses to exercise (‘central command’), and how it integrates with the muscle pressor reflex, has been sought. Animal experiments and functional imaging studies have provided clues, however, in humans the underlying electrophysiological activity of putative sites has never been measured until now. Functional neurosurgery has recently provided an opportunity to study these circuits directly. Emerging evidence suggests that the periaqueductal grey (PAG) could be a major site that integrates the cardiorespiratory response to exercise.1,2 We assessed the spectral changes in local field potentials (LFPs) recorded from deep brain nuclei in patients who had undergone neurosurgery for movement or pain disorders. We focused on the PAG, and compared it to other areas that might make up the sub-cortical circuitry, i.e. the internal globus pallidus (GPi), sub thalamic nucleus and thalamus. Patients with electrodes implanted were asked to perform light exercise on a cycle ergometer (study 1), or light handgrip exercise, followed by arterial occlusion of the exercised limb in order to maintain the pressor response (study 2). In the first study1, we re-visited the Krogh and Lindhard paradigm of 1913. Recordings were made during anticipation of exercise, actual exercise and recovery in an attempt to dissociate the exercise command from the movement itself. Anticipation of exercise resulted in increases in heart rate, arterial blood pressure and ventilation. The greatest neural changes were found in the doral lateral periaqueductal grey area (PAG) where anticipation of exercise was accompanied by an increase in activity. In the sub-thalamic nucleus there was a reduction during anticipation, but an increase with exercise. No significant changes were seen in the GPi during anticipation of exercise. In study 2, we re-visited Alam and Smirk’s paradigm of 1937. Here the dorsal lateral PAG also showed significant increases in neural activity during occlusion following exercise2. This period was associated with maintained elevated arterial blood pressure. Further increases in exercise intensity, and ischaemic exercise resulted in corresponding increases in PAG activity and ABP. No significant changes were seen in the activity of other nuclei during occlusion following exercise or during ischaemic exercise2. When all data are viewed together, we provide direct electrophysiological evidence highlighting the PAG as an important subcortical area in the neural circuitry of the cardiovascular response to exercise, since stimulation of this structure is known to increase arterial blood pressure in conscious humans 3.