Objectives: To review our management of infants discharged home receiving supplemental oxygen. Stable preterm infants receive low flow O-2 by nasal cannulae aiming for SaO(2) of greater than or equal to 95%. Oxygen-dependent infants must pass an air test (ability to maintain SaO(2) > 80% during 4 h disconnection from oxygen) before discharge home with supplemental oxygen. A sleep study is performed before nocturnal O-2 is ceased.Methods: Infants less than 33 weeks gestational age (GA) who were admitted January 1999-June 2001 and discharged home with supplemental oxygen were identified through the databases and medical records of the King Edward Memorial/Princess Margaret Hospitals. The data collected were compared with an audit performed a decade earlier.Results: Ninety-three infants were discharged home with supplemental oxygen between 1999 and 2001 (10% neonatal intensive care unit admissions less than 33 weeks GA; median GA 26 weeks (interquartile range 25-28). All infants had an air test before discharge: 63% failed the first air test and 30% at least two air tests. The median delay between the first air test and discharge was 2 weeks. The median postmenstrual age at discharge was 40 weeks gestation (interquartile range 38-41). Ninety infants had a sleep study before nocturnal oxygen was ceased and nine failed the first sleep study. Hospital readmission rate was 60%. More preterm infants (less than 33 weeks) were discharged with supplemental oxygen in 1999-2001 (10%, n = 96 in 1999-2001) than in 1987-1992 (2.5%, n = 53) and this was associated with an earlier discharge (40 vs 44 weeks postmenstrual age), lower oxygen requirements at discharge (60 vs 125 mL/min), earlier discontinuation of daytime and nocturnal oxygen (1 vs 4 months postmenstrual age and 2.5 vs 6 months postmenstrual age) and no increase in readmission rate (64% vs 60%). The incidence of bronchopulmonary dysplasia for these infants has remained stable at 20%.Conclusion: Our home oxygen programme, based on an air test predischarge and a sleep study prediscontinuation of nocturnal oxygen, facilitates early discharge home. Our data suggest that over the last decade, bronchopulmonary dysplasia is associated with less impairment in lung function. Further evidence from randomized clinical trials is required to determine optimal target range for oxygen saturation in preterm infants.