The effect of ambulatory oxygen use during pulmonary rehabilitation (PR) has not yet been clearly established, but many studies have shown benefit from oxygen acutely. Two small studies to date demonstrated no clear benefit when oxygen was used in patients who desaturated on exertion during a PR programme. One study showed the benefit of using oxygen during PR in patients who were normoxic at rest and desaturated to a minimum of 88% on exertion. We conducted a single-blind, randomised controlled trial comparing PR undertaken either with or without ambulatory oxygen, in those with demonstrable benefit from oxygen at baseline. Subjects from three PR services were recruited who, during baseline assessment, desaturated by more than 4% and to less than 90% on exertion, and walked 10% or more further with ambulatory oxygen on endurance shuttle walk test. Patients were randomised to either room air or oxygen via portable cylinder, titrated to optimise pulse oxygen saturation but up to a maximum flow rate of 6 L/min. All patients then completed a twice weekly, 6- to 7-week PR programme. Data were analysed as per protocol. Totally 51 patients completed the study. At the end of PR, both groups improved with PR, with patients in the oxygen group demonstrating a highly statistically significantly greater mean improvement in endurance walking distance than the controls, 490 m (95% confidence interval 228-750; p ≤ 0.001), as well as clinically, although not statistically, significant changes in quality of life. The use of ambulatory oxygen during a 6- to 7-week PR programme greatly improved endurance walking distance in patients who desaturated on exertion with a positive acute response to ambulatory oxygen at baseline.