We have studied, in six normal subjects, the effect of nitrous oxide sedation on the ventilatory pattern and oxygen saturation using pulse oximetry (SpO2) after hyperventilation to an end-tidal carbon dioxide partial pressure (PE'CO2) of 3 kPa. This value of PE'CO2 was shown to be less than the apnoeic threshold of all these subjects when their ventilation vs PE'CO2 response curves were plotted. All subjects became apnoeic when told to relax following hyperventilation while breathing 75% nitrous oxide for 90 s. Apnoea was defined as cessation of breathing for 20 s or more. The mean duration of apnoea was 78 s (range 29-130 s). All subjects demonstrated arterial desaturation (mean SpO2 75%, range 44-87%). In contrast, following hyperventilation with air, no apnoea was seen in any subject, although there was some evidence of desaturation (mean SpO2 92.5%, range 88-98%). It was concluded that subjects who are sedated with nitrous oxide behave similarly to those who are anaesthetized rather than to those who were fully conscious, in that they become apnoeic below the apnoeic threshold point. The reduction in SpO2 after hyperventilation was explained almost entirely by apnoea and may explain abnormalities of respiratory control and hypoxaemia in patients recovering from general anaesthesia or sedation accompanied by hypocapnia. This mechanism may be of importance in obstetric patients after breathing Entonox, when apnoea and hypoxaemia may reduce oxygen delivery to the fetus.