During wakefulness, obstructive sleep apnoea patients appear to compensate for an anatomically narrow upper airway by increasing upper airway dilator muscle activity, e.g. genioglossus, at least partly via a negative-pressure reflex that may be diminished in sleep. Previous studies have assessed the negative-pressure reflex using multi-unit, rectified, moving-time-average EMG recordings during brief pulses of negative upper-airway pressure. However, moving-time averaging probably obscures the true time-related reflex morphology, potentially masking transient excitatory and inhibitory components. This study aimed to re-examine the genioglossus negative-pressure reflex in detail, without moving-time averaging. Bipolar fine-wire electrodes were inserted per orally into the genioglossus muscle in 17 healthy subjects. Two upper airway pressure catheters were inserted per nasally. Genioglossus EMG reflex responses were generated via negative-pressure stimuli (approximately -10 cmH2O at the choanae, 250 ms duration) delivered during wakefulness and sleep. Ensemble-averaged, rectified, genioglossus EMG recordings demonstrated reflex activation (onset latency 26+/-1 ms; peak amplitude 231+/-29% of baseline) followed by a previously unreported suppression (peak latency 71+/-4 ms; 67+/-8% of baseline). Single-motor-unit activity, clearly identifiable in approximately 10% of trials in six subjects, showed a concomitant increase in the interspike interval from baseline (26+/-9 ms, P=0.01). Genioglossus negative-pressure reflex morphology and amplitude of the initial peak were maintained in non-rapid eye movement (NREM) sleep but suppression amplitude was more pronounced during NREM and declined further during REM sleep compared to wakefulness. These data indicate there are both excitatory and inhibitory components to the genioglossus negative-pressure reflex which are differentially affected by state.