To determine the impact of pulse oximeter monitoring on the incidence, severity and duration of hypoxaemia in the operating theatre (OT) and in the recovery room (RR), we investigated 200 patients in a randomized study. The extent of hypoxaemia in the OT was compared with that in the RR. Adult inpatients were allocated randomly to two groups: group I, pulse oximeter data and alarms "available"; group II, these data "unavailable" to the anaesthesia team and RR staff. Hypoxaemia was graded into four values of oxyhaemoglobin saturation (SpO2). The incidence of hypoxaemia was reduced significantly in group I in both OT and RR. In the OT, five patients in group II suffered SpO2 less than 76% compared with none in group I (P less than 0.02). In group II in the RR, seven patients suffered SpO2 less than 81%; three of these had SpO2 less than 76%. No patients in group I exhibited such small values of saturation. The smallest recorded SpO2 in the OT and the RR was significantly greater in group I. The cumulative duration of hypoxaemia was significantly less in group I in the RR, but not in the OT. The incidence and severity of hypoxaemia in the OT and in the RR were comparable, whereas the cumulative duration of hypoxaemia was significantly greater in the RR than in the OT. The occurrence of hypoxaemia in an individual patient in the OT significantly increased this patient's risk of suffering hypoxaemia in the RR. We conclude that the extent of hypoxaemia, especially in the RR, may be reduced significantly by pulse oximeter monitoring, but even with the information provided, some patients still develop hypoxaemia.