1. We investigated the possibility of a cortical contribution to human respiration by recording from the scalp of awake subjects the premotor cerebral potentials that are known to precede voluntary limb movements. 2. Electroencephalographic activity (EEG) was recorded from scalp electrodes and averaged for 1.8-2.0 s before the time at which airway pressure exceeded an inspiratory or expiratory threshold. Clear premotor cerebral potentials were recorded during brisk, self-paced nasal inhalations or exhalations. In ten subjects, a slow cortical negativity (Bereitschaftspotential) was apparent in the averaged EEG, commencing 1.2 +/- 0.3 s before the onset of inspiratory (scalene) or expiratory (abdominal) muscle activity (EMG). It was maximal at the vertex, with a mean slope of 12.3 +/- 5.8 microV/s, and was followed by a post-movement positivity. 3. In four subjects the inspiratory premotor potential culminated in a large negativity, the motor potential, which began 24 +/- 15 ms before the onset of scalene EMG. It is argued that such a short latency is consistent with a volitionally generated respiratory command which travels relatively directly to the respiratory muscles, having a total central delay which is no longer than that for voluntary finger movements. 4. That the respiratory premotor and motor potentials did not originate in subcortical structures was supported by their absence in a patient suffering from chronic reflexogenic hiccups, in whom cerebral activity was back-averaged from each brisk hiccup. 5. During quiet breathing, in which subjects were relaxed and distracted from thinking about their respiration, no premotor cerebral potentials preceding inspiration could be detected. This failure was not due to the slow rate of rise of inspiratory activity during quiet breathing as compared with a brisk sniff, because premotor potentials were detected when subjects intermittently generated slow active expiratory efforts. 6. These observations suggest that during quiet breathing the cerebral cortex does not contribute to respiratory drive on a breath-by-breath basis. Conversely, the presence of clear premotor cerebral potentials when subjects performed self-paced inspiratory or expiratory manoeuvres illustrates the powerful cortical projection to human respiratory muscles.