To assess the effects of selective sleep loss on ventilation during recovery sleep, we deprived 10 healthy young adult humans of rapid-eye-movement (REM) sleep for 48 h and compared ventilation measured during the recovery night with that measured during the baseline night. At a later date we repeated the study using awakenings during non-rapid-eye-movement (NREM) sleep at the same frequency as in REM sleep deprivation. Neither intervention produced significant changes in average minute ventilation during presleep wakefulness, NREM sleep, or the first REM sleep period. By contrast, both interventions resulted in an increased frequency of breaths, in which ventilation was reduced below the range for tonic REM sleep, and in an increased number of longer episodes, in which ventilation was reduced during the first REM sleep period on the recovery night. The changes after REM sleep deprivation were largely due to an increase in the duration of the REM sleep period with an increase in the total phasic activity and, to a lesser extent, to changes in the relationship between ventilatory components and phasic eye movements. The changes in ventilation after partial NREM sleep deprivation were associated with more pronounced changes in the relationship between specific ventilatory components and eye movement density, whereas no change was observed in the composition of the first REM sleep period. These findings demonstrate that sleep deprivation leads to changes in ventilation during subsequent REM sleep.