The need for oxygen increases with activity in patients with COPD and on long-term oxygen treatment (LTOT), leading to periods of hypoxemia, which may influence the patient’s performance. This study aimed to evaluate the effect of automated oxygen titration compared to usual fixed-dose oxygen treatment during walking on dyspnea and endurance in patients with COPD and on LTOT. In a double-blinded randomised crossover trial, 33 patients were assigned to use either automated oxygen titration or the usual fixed-dose in a random order in two walking tests. A closed-loop device, O2matic delivered a variable oxygen dose set with a target saturation of 90–94%. The patients had a home oxygen flow of (mean ± SD) 1.6 ± 0.9 L/min. At the last corresponding isotime in the endurance shuttle walk test, the patients reported dyspnea equal to median (IQR) 4 (3–6) when using automated oxygen titration and 8 (5–9) when using fixed doses, p < 0.001. The patients walked 10.9 (6.5–14.9) min with automated oxygen compared to 5.5 (3.3–7.9) min with fixed-dose, p < 0.001. Walking with automated oxygen titration had a statistically significant and clinically important effect on dyspnea. Furthermore, the patients walked for a 98% longer time when hypoxemia was reduced with a more well-matched, personalised oxygen treatment.