Several investigators have observed that irregular breathing occurs during rapid-eye-movement (REM) sleep in healthy subjects, with ventilatory suppression being prominent during active eye movements [phasic REM (PREM) sleep] as opposed to tonic REM (TREM) sleep, when ocular activity is absent and ventilation more regular. Inasmuch as considerable data suggest that rapid eye movements are a manifestation of sleep-induced neural events that may importantly influence respiratory neurons, we hypothesized that upper airway dilator muscle activation may also be suppressed during periods of active eye movements in REM sleep. We studied six normal men during single nocturnal sleep studies. Standard sleep-staging parameters, ventilation, and genioglossus and alae nasi electromyograms (EMG) were continuously recorded during the study. There were no significant differences in minute ventilation, tidal volume, or any index of genioglossus or alae nasi EMG amplitude between non-REM (NREM) and REM sleep, when REM was analyzed as a single sleep stage. Each breath during REM sleep was scored as "phasic" or "tonic," depending on its proximity to REM deflections on the electrooculogram. Comparison of all three sleep states (NREM, PREM, and TREM) revealed that peak inspiratory genioglossus and alae nasi EMG activities were significantly decreased during PREM sleep compared with TREM sleep [genioglossus (arbitrary units): NREM 49 +/- 12 (mean +/- SE), TREM 49 +/- 5, PREM 20 +/- 5 (P less than 0.05, PREM different from TREM and NREM); alae nasi: NREM 16 +/- 4, TREM 38 +/- 7, PREM 10 +/- 4 (P less than 0.05, PREM different from TREM)]. We also observed, as have others, that ventilation, tidal volume, and mean inspiratory airflow were significantly decreased and respiratory frequency was increased during PREM sleep compared with both TREM and NREM sleep. We conclude that hypoventilation occurs in concert with reduced upper airway dilator muscle activation during PREM sleep by mechanisms that remain to be established.