In Study A, the incidence of arterial oxygen desaturation was studied using pulse oximetry (SaO2) in 100 sedated and 100 nonsedated patients breathing room air who underwent diagnostic upper gastrointestinal endoscopy. Hypoxia (SaO2 92% or less of at least 15 s duration) occurred in 17% and 6% of sedated patients and nonsedated patients, respectively (p < 0.03). Mild desaturation (SaO2 94% or less and less than 15 s duration) occurred in 47% of sedated patients compared with 12% of nonsedated patients (p < 0.001). In Study B, the effects of supplementary oxygen therapy and the effects of different pre-oxygenation times on arterial oxygen saturation (SaO2) in sedated patients were studied using pulse oximetry. One hundred and twenty patients who underwent diagnostic upper gastrointestinal endoscopy with intravenous sedation were studied. Patients were randomly allocated to one of four groups: Group A (n = 30) received no supplementary oxygen while Groups B-D received supplementary oxygen at 4 1 x min(-1) via nasal cannulae. The pre-oxygenation time in Group B (n = 30) was zero minutes, Group C (n = 30) was 2 min and Group D (n = 30) was 5 min before sedation and introduction of the endoscope. Hypoxia occurred in seven of the 30 patients in Group A and none in groups B, C and D (p < 0.001). We conclude that desaturation and hypoxia is common in patients undergoing upper gastrointestinal endoscopy with and without sedation. Sedation significantly increases the incidence of desaturation and hypoxia. Supplementary nasal oxygen at 4 1 x min(-1) in sedated patients abolishes desaturation and hypoxia. Pre-oxygenation confers no additional benefit.