Objective To compare the effect of two different automated oxygen control devices on target range (TR) time and occurrence of hypoxaemic and hyperoxaemic episodes. Design Randomised cross-over study.