We evaluated an auto-nasal continuous positive airway pressure (nCPAP) prototype (MC+; SEFAM, Nancy, France) in which apnoea/hypopnoea detection was disabled and nasal mask pressure vibration detection was the only mode of pressure setting. The device was tested in 15 previously untreated obstructive sleep apnoea patients during a night with polysomnography. We observed that a single night of auto-nCPAP improved the apnoea/hypnoea index (AHI) (12 +/- 21 vs 51 +/- 31 disordered breathing events.h-1 of sleep (mean +/- SD)), the awakening-arousal index (13 +/- 20 vs 40 +/- 26 arousals.h-1 of sleep), and duration of slow wave sleep (102 +/- 49 vs 71 +/- 56 min) but not of rapid eye movement (REM) sleep (55 +/- 31 vs 64 +/- 38 min). Auto-nCPAP was effective (apnoea/hypopnoea and arousal indices < 10 events.h-1) in all but three patients. Auto-nCPAP was ineffective in one patient, whose obstructive respiratory events were not preceded by nasal mask pressure vibration detection, and in two patients who were quasi-permanent mouth breathers. Snoring detection may be effective in sleep apnoea syndrome with heavy snoring and without permanent mouth breathing, during the first night of nasal continuous positive airway pressure treatment.