Pneumotachographic recording of ventilation requires an air-tight nasobuccal face-mask which might disturb the subjects' sleep and modify their breathing patterns. In a cross-over randomized study, we compared recordings obtained with and without a face-mask in 20 healthy normal volunteers (10 males, 10 females, mean age: 21.7 +/- SEM 0.5 years). The sleep modifications that resulted from the presence of the face-mask were a lengthening of the REM-sleep latency (+ 38%, p less than 0.01), an increase in intrasleep waking-time (+ 55%, p less than 0.05), an increase in sleep stage-1 duration (+ 40%, p less than 0.02), and a decrease in REM-sleep duration (-20%, p less than 0.01), resulting in a decrease in total sleep duration (-9%, p less than 0.05). The sleep-latency, the number of awakenings, the sleep stage-2 duration, the deep slow wave sleep duration, the sleep stability and the number of sleep cycles were not significantly modified. The respiratory frequency was unchanged in any of the sleep stages. No obstructive or mixed apnoea was observed during the nights when the subjects wore the face-mask. The apnoea index for central apnoeas was low and remained unchanged with and without the face-mask (0.30 +/- 0.10 and 0.35 +/- 0.10 respectively). Thus the observed modifications do not rule out the basic advantage of the pneumotachographic method, that of being the only method to-date providing a direct quantitative measurement of respiratory flow and flow rate during sleep.