To investigate the response of inspiratory and expiratory muscles to naturally occurring inspiratory resistive loads in the absence of conscious control, five male "snorers" were studied during non-rapid-eye-movement (NREM) sleep with and without continuous positive airway pressure (CPAP). Diaphragm (EMGdi) and scalene (EMGsc) electromyographic activity were monitored with surface electrodes and abdominal EMG activity (EMGab) with wire electrodes. Subjects were studied in the following conditions: 1) awake, 2) stage 2 sleep, 3) stage 3/4 sleep, 4) CPAP during stage 3/4 sleep, 5) CPAP plus end-tidal CO2 pressure (PETCO2) isocapnic to stage 2 sleep, and 6) CPAP plus PETCO2 isocapnic to stage 3/4 sleep. Inspired pulmonary resistance (RL) at peak flow rate and PETCO2 increased in all stages of sleep. Activity of EMGdi, EMGsc, and EMGab increased significantly in stage 3/4 sleep. CPAP reduced RL at peak flow, increased tidal volume and expired ventilation, and reduced PETCO2. EMGdi and EMGsc were reduced, and EMGab was silenced. During CPAP, with CO2 added to make PETCO2 isocapnic to stage 3/4 sleep, EMGsc and EMGab increased, but EMGdi was augmented in only one-half of the trials. EMG activity in this condition, however, was only 75% (EMGsc) and 43% (EMGab) of the activity observed during eupneic breathing in stage 3/4 sleep when PETCO2 was equal but RL was much higher. We conclude that during NREM sleep 1) inspiratory and expiratory muscles respond to internal inspiratory resistive loads and the associated dynamic airway narrowing and turbulent flow developed throughout inspiration, 2) some of the augmentation of respiratory muscle activity is also due to the hypercapnia that accompanies loading, and 3) the abdominal muscles are the most sensitive to load and CO2 and the diaphragm is the least sensitive.